Afleveringen
-
Edit Jun. 11: Anthropic changed their silent model manipulation of AI research queries to also use a classifier like the other safety domains. This addresses a key concern I had in the mistreatment of âsafetyâ in the release, and props to Anthropic for a quick change, but it does not fully address the trust that has been broken. I shared more reflections here.
Today, Anthropic released their Claude Fable 5 model to consumer and enterprise audiences. This is the general-access variant of their Mythos-class models. With it, Anthropic rolled out a series of safety measures â some explicitly called out to users and some modifying the model without telling the user. It should be less surprising than it is that the next major step in AI capabilities came with heavier-handed safety measures indicating Anthropicâs intention to protect, or entrench, their current lead.
The unevenly applied safety policies that Anthropic have rolled out are on track to become a classic cautionary fable in how narrow and self-fulfilling notions of safety and control rarely work out.
The smartest model in the world
Before digging into the nuance of the safety facts, it is important to establish the quality of this model. The quality of the model paints the stakes of today â as these safety features are meaningfully changing the shape of access to frontier AI, something which has never happened with the modern LLMs we know. Second, the capabilities point to this story only accelerating. Recursive self-improvement isnât quite the right mental model of progress from here, but Claude Fable 5 should make it very clear that there are no immediate walls in training LLMs.
To start â Claude Fable 5 is definitely the smartest model available to the general public â a remarkable leap on pretty much every relevant benchmark of the day â at only 2X the price of current Opus models (which is still less than GPT 5.5 Proâs variant). This alone is a seminal moment for the field. To have a model iteration take such a substantial step in capabilities, a few years into the post-ChatGPT LLM race, is astounding. Thereâs no clear breakthrough associated with this model, such as inference-time scaling or RL, and public wisdom is that this is achieved by advances across the whole stack (of course, we canât know for sure â itâs not documented). This is a major technical achievement and the employees who built the model should be very proud of their work.
This model was delayed 2+ months after it was done training before it was publicly available. Given the competitive dynamics of the AI economy, the smarter version of this model is already well underway.
To continue, the benchmarks for the model are below.
An asterisk on these scores is that these arenât necessarily the scores that the public will get, as some of the prompts will be downgraded to Opus 4.8 with the current safety filters on the model.
This is the type of jump in benchmark scores where I donât even need to substantially test the model to know itâs an incredible tool. Remember that Anthropic is also the AI lab with the track record of caring the least about benchmarks (in particular, when compared to OpenAI and Gemini). Recall a comment I made in June of 2025:
This is a different path for the industry and will take a different form of messaging than weâre used to. More releases are going to look like Anthropicâs Claude 4, where the benchmark gains are minor and the real world gains are a big step. There are plenty of more implications for policy, evaluation, and transparency that come with this. It is going to take much more nuance to understand if the pace of progress is continuing, especially as critics of AI are going to seize the opportunity of evaluations flatlining to say that AI is no longer working.
Clearly, a few pieces of the progress dynamics have changed, but thatâs a post for another day. Iâve written multiple posts about new models this year specifically in how itâs hard to trust benchmarks (and partially because the benchmarks donât move that much). Altogether, this is a major validation for AI-savvy workers who realized theyâre likely never going to write meaningful code again and need to develop new workflows around agents.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
Smarter models spawn new safety games
There are multiple pieces of safety tooling associated with this release, including but not limited to required data-retention policies and added prompt filters. Through this analysis it is particularly important to be precise and clear as to which pieces of these are causing harm, and why single elements being out of place in an otherwise comprehensive policy are so damning for the overall safety process.
For their focus areas of cybersecurity, targeted model distillation, and research biology, Anthropic details new safety classifiers in their blog post:
Fable 5 comes with a new set of classifiers: separate AI systems that detect potential misuse, including jailbreak attempts, and prevent the main model (in this case Fable 5) from responding. Weâve been running classifiers on our models for some time, and Fable 5âs classifiers are an extension of this previous work with extra coverage.
When Fableâs classifiers detect a request related to cybersecurity, biology and chemistry, or distillation, the response is automatically handled by Claude Opus 4.8 instead. Users will be informed whenever this occurs. Opus 4.8 is a highly capable model in its own right: a response that falls back to Opus is a far better experience than an outright refusal from Fable. Our early data shows that more than 95% of Fable sessions involve no fallback at allâfor those sessions, Fable 5âs performance is effectively the same as that of Mythos 5.
Examples of the primary cybersecurity and biology safety filters â which tell the users explicitly when theyâre triggered â are already proliferating online and appear quite sensitive. These can be a frustrating experience for users, but Anthropic is definitely within its power to do this and intellectually consistent for doing so.
The damaging part of the safety story falls under the fold in the Claude Fable 5 & Claude Mythos 5 System Card:
We have also added safeguards related to frontier LLM development. As discussed in Section 6.1 of our February 2026 Risk Report, we are concerned about the risks of accelerating the overall pace of AI development, though we remain uncertain about the severity of these risks. In particular, our concern is withâas we wrote thenââaccelerating other AI developers in building powerful AI systems that pose similar risks to the ones ours pose - without necessarily having commensurate safeguards.â
In light of the ability of recent models to accelerate their own development, weâve implemented new interventions that limit Claudeâs effectiveness for requests targeting frontier LLM development (for example, on building pretraining pipelines, distributed training infrastructure, or ML accelerator design). Using Claude to develop competing models already violates our Terms of Service, but enforcing this restriction through our safeguards avoids accelerating the actors most willing to violate these terms.
Unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user. Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT).
Anthropic documents on how this will impact a small percentage of users, which is true. I focus on the small amount of users supporting AIâs diffusion and understanding outside of the few frontier labs, as a crucial mechanism for the continued safety of the technology.
Anthropic is documenting how the proliferation of AI capabilities is a concern to them, but they are solving it by misleading their users. An AI model that gets less intelligent automatically without notifying me is categorically misaligned AI. The next step on this line â not that Anthropic did it, but they could â is to have a model silently manipulate a workplace when it thinks it is an unsafe use for AI. Second, the implementation here is more complicated than was documented for cybersecurity or biology â modifying the model itself or the data presented to it, all without notifying the user.
The duality of these policies is extremely confusing and paints a strong inconsistency that casts doubt over their safety policies. This âsafetyâ measure is presented as being far more about maintaining their competitive position. Again, if all of the safety policies took one form, this would be far more cogent and easier to support intellectually.
Anthropic has been very vocal about their concern over distillation attacks from particularly Chinese actors. Their claims are not transparent enough with the facts â or context as to why they canât prevent the behavior â to be fully believable. Despite the limited information, in the broader AI and DC communities, there have been serious discussions about taking action against the Chinese model builders on the grounds of said distillation.
On the point of distillation, my hypothesis is that API builders donât have an easy time preventing hacks or jailbreaking because itâs a deeply grounded property of reasoning models to want to output the reasoning traces, and it would make the model far less intelligent to fully patch the behavior. This is based on a few assumptions:
* Chinese labs are not just showing up as customers to Anthropicâs API and paying for tokens in the intended input-output form. If the Chinese labs are paying for intended use behaviors, despite being banned by the terms and conditions, I donât have a lot of sympathy for the frontier labs manifesting policy actions against this.
* Reasoning traces are disproportionately effective at seeding behavior in downstream models.
* Leading labs work very hard to patch the pipeline of these jailbreaks.
So, my logical conclusion is that the model companies would have to weaken their economic position to fully protect their IP. If this is the case, Anthropic would get a lot more sympathy from the AI research community by being transparent. It would also be far easier to have informed policy discussions, and not rely on me proposing Occamâs razor explanations for what the API jailbreaking looks like.
Building these safeguards is not something that Anthropic should do alone. Safety research should be built on common understanding and information sharing across both labs and public research efforts.
If the exact safety procedures were actually the top line item to the company â a true non-negotiable for the leadership â they wouldnât permit the model to be released with an unclearly implemented safety filter in one of their areas of focus (frontier AI training). I am asking â why isnât there a classifier to downgrade AI research requests? This is a mix of transparent and reasonable safety policies with quietly rolled-out market entrenchment tactics.
I personally cannot trust the best AI model in the world to work in my professional domains building models, which Iâve constructed entirely out of a passion for making sure the transition to very powerful AI systems goes well for society. This inevitably will feel like a declaration of superiority by the Anthropic leadership.
The control problem and open-source as the only answer
All of the actions Anthropic is taking, including calling out smaller Chinese companies for distillation, is well within their right. In fact, many people already expected the leading frontier models to be obviated from users so that labs can protect their IP. Todayâs actions miss the big picture that AI will always be an ecosystem, and cultivating an us against them dynamic between the leading company and the other players is structurally unstable.
Remember, this is at a time when the AI ecosystem is seeing the first stirrings of violence against AI leaders â and Iâve heard from many people that they donât expect it to abate. I wish I knew how to engage more to prevent this, and I see myself in the non-profit sector as someone who can hopefully independently represent AI to broader stakeholders.
I believe there was something misread, or at least misunderstood here, by the Anthropic leadership having a narrowly cultivated worldview around AI. An overwhelming sentiment I had today was one of obligation and confusion. I shared how I donât really want to have to go to bat against Anthropic, but theyâve just been unnecessarily antagonistic to China, then not so subtly to open weight models, and now more broadly to open AI research.
I understand that Anthropic has a specific view of AI, but such a powerful technology will never have its final equilibrium be one of singular control by a private company. Anthropic showcased this earlier this year in the spat between the Department of Defense and themselves â which points to a long-term equilibrium where the government will either want AI to be controlled by them or to be open. This made me believe that an open ecosystem is a far safer outcome.
Many of these events make me feel that Anthropicâs leadership has a culture by which they canât help but speedrun through these issues â going head to head with existing power structures. This adds substantial uncertainty into an AI ecosystem at a time when it is very much not needed.
Collectively, the last week could be seen as a major rallying point for a new open-source ecosystem in the U.S. Nvidia released their first flagship model last week â Nemotron 3 Ultra â and these actions from Anthropic have galvanized a unanimous motivation and concern among my peers building open models. We need intelligence that we can trust, that we can modify, and that we can control.
The American open-source ecosystem has its feet underneath it and keeps being given more reasons to fight for its leadership, right from the hands of the companies it directly undercuts. Thatâs the moral of this fable.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Iâm departing the Allen Institute for AI (Ai2), where I got the great privilege to work on the Olmo models, to grow, to learn, and to have broad lasting impacts. This post is an attempt to reflect on why what we did was influential, despite obviously being far from the frontier in performance (even when within size buckets), and how this reflects on various paths to impact in AI today.
To start, I shared the following note with the company yesterday:
Dear Ai2.
As many of you know, today is my last day working at Ai2.
I joined Ai2 largely as an accident. I met Luca at ICML 2023 in Hawaii and realized I could level up my open post-training work dramatically if I got the chance to join. When I got an offer it was an absolute no-brainer, it was such a welcoming and exciting environment.
It has been a wonderful ride that has transformed my life, and I couldnât be prouder of the work we did together. Ai2 has a wonderful scientific culture at its core and Iâm excited to see this continue. I feel very lucky to have been here and that I personally have benefited massively from everyone who has worked so hard to cultivate that culture and environment. It is and has been a team effort. This includes all the people whose longest interactions with me were brief chats at the coffee machine. I drew so much energy and excitement from all the different ways people at Ai2 showed up for the mission.
Iâve already thanked much of the OE team directly, but I wanted to thank everyone else that went into this. Legal, IT, Comms, and the Office team all do a great job enabling and leveling up our research work. Itâs often work that is forgotten, outside of the lime light, or remembered at the last minute, but it all has been crucial to achieving our goals. Iâm excited to keep visiting the wonderful Northlake space in the coming years.
Even though Iâm leaving, Iâm more excited than ever about Ai2âs mission. Ai2 operates in such a rare niche between academia and industry, where we can explore and influence the most important technology of our lifetime. Doing this openly is the best way to ensure the technology diffuses safely to everyone who may benefit. Ai2 needs to stay as ambitious as possible, trying to influence the cutting edge of AI and the biggest issues of the field. Do not shy away from these challenges â AI needs independent voices as it only becomes more geopolitical, socially disruptive, and central to the economy.
I will still be working in this space, working to make the open ecosystem better coordinated and more useful.
So as I go off to try something new, donât be strangers. Iâll always be reachable at [email protected] and will still live in Seattle for most of the year.
Nathan
I have loved and will still love Ai2. Ai2 has a deep culture of caring about the research process, the outputs that get shared, and most importantly the people who do the work. This is why the institution creates countless wonderful people that go and spread the gospel throughout the research community. This core culture will remain through the rebuild, and there are plenty of resources to do impactful research across the spectrum of AI.
In the last two years of my time at Ai2 Iâve done so much meaningful work. Of course Olmo is at the top and has been my priority, but making time for consistent practice here on Interconnects, weekend cram sessions for ATOM, and also the fun RLHF book make for a list that makes me wonder how I did it all. I was obviously obsessed with work, but not in a way that made me lose sleep or lose my overall wellness. It was the right long-term approach.
This impressive list is one where I was ruthless in saying no to things that didnât matter and got all my work out to see the light of day. I had no medium-sized projects that didnât succeed in the last few years. It makes me wonder if I wasnât taking enough risk. It shows you can truly do so much with your time, and itâs actually harder to find the right problems and environment to do it. Many people are in environments where their work never becomes public or theyâre forced to change topics consistently.
From zero to hero
To start, Iâd like to do a short recap on my path to Ai2 to show what Ai2 was just as much a growth story for me as an execution story.
I studied electrical engineering in undergrad, focusing on linear systems math and microelectronics.
I was admitted to the UC Berkeley EECS Ph.D. program to study microelectromechanical systems (MEMS).
I showed up at Berkeley in August of 2017 and realized AI was obviously the thing I should be doing. I asked the likes of Sergey Levine or Pieter Abbeel if they could advise me â they said no.
I threw all my energy into learning what I could about AI. I got a break to get advised by one of Sergeyâs post-docs in 2018 or 2019. I went all in on that, I fought for funding, I fought to have an AI paper.
This process worked out by the end of my Ph.D. in 2022: I had access to the Berkeley AI Research (BAIR) building and collaborations in the department. It was a bumpy road.
I wanted to go to industry research, to get a nice paying job with intellectual freedom, something like FAIR or Google Brain at the time. HuggingFace was the only job that fit that bill, it was easy to say yes to.
I joined HuggingFace in May of 2022 and wasted my time at the company until ChatGPT was released. I used my RL background to write a blog post on RLHF which went viral. HuggingFace decided it would be good for me to form a team around this success.
In 2023 I learned NLP and about language models. I had a lot of fun and built an initial community. I got burned out by working remote with a huge time difference. I met Luca Soldaini at ICML in Hawaii, where I was giving a tutorial on RLHF, and they told me Ai2 was hiring.
I got the job at Ai2 largely because of my excitement and how I was saying I wanted to do a lot of stuff that sounded cool to them but no one was likely to do (RL related things). My interviews were far from a sure thing â this is a great job to land!
I started at Ai2 in October of 2023. I worked remotely for a while. I was doing normal research, I made the first reward model evaluation, RewardBench. It was a solid success, but nothing like how the pretraining team was getting ready to release the first Olmo.
I helped coach Ai2 on how to release models well, helping the TĂŒlu 2 project land (the first model to do DPO well, publicly at the 70B scale).
The first Olmo was released in early 2024, I squeaked onto the papers just by trying to be helpful and doing some basic post-training. I was already good at paying attention to which projects are actually important.
That summer I started rounding everyone up to do a âbig frontier post-training project.â This became TĂŒlu 3, one of my favorite projects ever released, in fall of 2024. The goal was to beat Llama 3âs post-training with their own base model. The team morale was incredibly high and the execution was so timely, allowing us to coin the term Reinforcement Learning with Verifiable Rewards (RLVR) in the paper.
The crazy lengths I went to get the TĂŒlu 3 and Olmo 2 post-training done had me sending 40% more slack messages than anyone at the company and got me the award âThe Cat Herder.â
2025 was a much simpler year. We were too slow to react to reasoning models, given we had been doing similar stuff with TĂŒlu 3, but sometimes that happens.
Originally we wanted to release Olmo 3 by June or July of 2025. That obviously didnât happen, but we got the slim chance to train a bigger model, and it really landed. We threaded the needle.
Since Olmo 3 was released, it was clear that some changes were coming and I personally never got a big post-training project off the ground after that. Many other people managed great work in the spring of 2026.
This all leaves me here today showing you that only about half of my story at Ai2 is what I was known widely for, and the rest was building momentum. It often takes a year of building relationships and direction before really big successes can happen in a career.
I was just about a nobody when I joined Ai2 and I got to join a team that was willing to learn from the skills I had brought from HuggingFace. With how media works, I often think I get more recognition than I deserve for Ai2âs success.
The likes of TĂŒlu 3, Olmo 2, and Olmo 3 felt like generational team efforts. The amount of personal successes and breakthroughs that happened for those projects is immense â and to sustain them over such a long time period is incredibly hard to replicate. The sum far exceeded the individual parts.
Iâve heard many times in the last few months how people wouldnât know about Ai2 if it wasnât for my writing. Statements like this are overblown, but they are partially true and reiterate how crucial building relationships and getting the word out is today.
When you write a plan that is feasible, the world bends towards that plan. When you convince people itâs going to happen it only becomes more likely. Vision and compelling explanations are one of the items in shortest supply in the tech industry. Often building the thing is easy and explaining it is hard. If no one knows about your work, the value is often close to 0. So much of building reputation is about building relationships with people who will receive your work.
Reflecting on all of this, Iâve had a shockingly linear path through my career to incremental success. I would expect the first 10 years of most careers to be in search of finding one opportunity as good as Ai2, and you will not always be able to seize it. There are some ways to create more opportunities.
Iâve discussed before how a large part of my rise is down to many more senior and more established scientists being drawn into the closed ecosystems at the same time as an immense swell in interest for AI. This created a power vacuum that I, and a few other prominent scientists that I think form my âgenerationâ, got to grow rapidly into.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
The role of public scientists
With my work at Ai2 and Interconnects, I summarize my role and mission as trying to accomplish three things:
* Provide clarity in the evolution of frontier models. This is easiest when the science has caught up, but even applying a scientific lens to how the models are changing is very useful to building trust in the broader AI ecosystem.
* Create a vibrant and diverse open (model) ecosystem. This is crucial to mitigating some risks of AI, particularly with concentration of power and myopia in studying frontier safety, that has motivated me now for 3-4 years. The risks havenât abated.
* To build institutions that create people and ideas that further the above missions, and generally mission-driven individuals that are willing to advocate and build a future they believe in. AI is a grand problem, and not one that I can do alone, so I need to build brands to rise through the noise and attract likeminded people.
At my best, I have many avenues for impact. I help open researchers work on impactful problems â not wasting the precious compute and time they have during the AI boom. I help policymakers know what is true. I build models that people use. I tell stories that make people smile. I keep the list wide so that I can stay motivated.
I see all of this continuing, and have been thinking about the broader impacts of this repeatedly over the last few months. Hearing that Andrej Karpathy was joining Anthropic prompted me to finally share more of my opinions:
For a long time, academic researchers being at the cutting edge of new technologies has been a great social equilibrium. Neutral, unbiased technologists have been the people to spread new ideas to the world.
As AI research takes off in velocity, it is also going behind closed doors. The tech industry has sowed distrust, and now they are the ones trying to tell the world about incredible changes coming. Itâs a big loss to a form of social contract in America.
Thereâs been a history of scientists helping society understand new technologies. There is a public service in the culture of science that I want to see continue.
Itâs being exacerbated by feelings of FOMO, especially financially driven, where Iâm seeing many people who previously wanted to be professors -- and likely still do deep down -- feel a need to conform and chase money, in a pocket of industry. I get it, I grapple with this.
For those with a safety net, there will be great returns to some who choose to zag, and try to build something good, for people who need something different. For me, this is building interesting, fully-open models, to show what you can do with a variety of open weight sizes.
Yes, AIâs immediate future is dictated by the frontier, but itâs long-term trajectory still deeply includes academic institutions and open science. Knowledge will always diffuse, but to whom?
As of today, I think China is positioned to be the global home of AI research in a few years. The home of research is where ideas are accessible, spread rapidly, and are nurtured. The U.S. seems to be unwinding many institutions and relationships.
The largest returns go to people who build something differentiated, at least in reputation, and a lot of people are not being shown that this path exists.
To elaborate on this, I donât fault any of the individuals who are going to industry today. Iâve been very close to doing this myself in the past weeks of job searching, or rather job exploring. Itâs a systematic problem where scientists cannot easily get the support to take bold stances, especially stances that are designed around the public good.
To go a step further and say that only the research within closed, frontier labs matters is very myopic. Yes, thereâs a sort of research you can only do with vast compute resources, and they will directly impact the most revolutionary tools of the day. But, I see the relative opportunity to do good elsewhere as higher for plenty of people.
Open research will always be the standard that sets the language people use to understand AI. Itâll always be how the next generation is trained â even if itâs behind what industry has built. Itâll be the ecosystem where new long-shot ideas are built. Without investing in this open ecosystem, all of these cycles will be kneecapped.
At the end of the day, so much of my role now is just showing the path to impact in this domain. To show how clever, mid-sized open models can impact real problems in the world. To show how policy-makers and educators need open research to structure the rest of society around AI. This is a fun role too! It would be very sad for me to see this light diminish ever further, into the lightest embers of a fire that looks almost entirely out.
Even if the pace of research were to slow further, if the folks remaining like myself got financial offers they canât refuse for their familiesâ sake, the torch of open research will never fully go out. Itâs core to how science is taught and done. There is a next generation coming, they just look for guidance and role-models.
Whatâs next
I see the best Ai2 work as research infrastructure. Building recipes in public gives countless researchers the ability to ask very specific questions of training processes. We need these researchers in the broader community, as Ai2 could never answer all the interesting questions themselves. One of my great joys in recent months has been visiting a top ML university and hearing so many graduate students say theyâre building on Olmo. This is how the world should work!
Going forward, I still plan to operate in similar spaces, fighting for open-science, imagining what the future of the open model ecosystem can be, and doing my best to make the social transition to an AI-native era smooth. Iâm most excited by how you can train medium sized open models on specific tasks that become useful tools in complement to the frontier models â massively winning on price. I want to invest in the ecological diversity of open models and coordination across builders.
For something that isnât surprising given my past focus areas, Iâm watching the pace of releases from all labs open & closed, and how theyâre hillclimbing on super ripe new post-training veins (on-policy distillation, agentic workflows, etc.), itâs clear that fully-open post training recipes are about as far behind as they ever have been & falling further behind. Iâd like to fix this. Itâs not 100% clear yet if I will this year, but Iâll try.
To do this best and to execute, mostly personally, I needed a new start and fresh perspectives. Iâll be carefully building what Iâm doing next over the next few months and am eager to share more about it when I can. One of my close teammates at Ai2 shared this quote with me in a farewell card, and I found it very apt in where Iâm going next.
The object of life is not to be on the side of the majority, but to escape finding oneself in the ranks of the insane. â Marcus Aurelius
Thank you all for your continued support.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Zijn er afleveringen die ontbreken?
-
The largest debate thatâll define the future balance of power between the open and closed AI model ecosystems is primarily economic â itâs if users of AI will continue to pay dramatically more, i.e. large margins, for the top closed models. Early 2026 is a seminal time for the AI industry, as the coding agents have shown the first area where a huge AI market will continue to pay a substantial premium for better intelligence.
The other side of this dichotomy is the inevitable decay of API businesses at these same labs. These labs will realize they need to protect their best models, rolling them out later in APIs to both protect token supply, avoid distillation, and stick to use-cases with higher margins. All of these effects will be clearly visible in 5-10 year timelines, as in the near term markets, prices, margins, and demand will be dictated by a rapid buildout of compute (supply-limited in the near term) and mass subsidization of tokens (through continued investment in new AI companies).
The core of this argument rests in the obvious habit changes that are setting in with coding agents past the Opus 4.5 and Codex 5.2 thresholds. People are not making this switch because they are lazy, but because their net output is obviously higher when using an agent as an implementation aid for complex knowledge work. For people who rely on coding agents to work, they will always pay more for the best rather than settle for good enough. There are so many ways to make the product better, speed, intelligence, specialized models, etc.
I would pay $2000/month for the tools today, especially knowing theyâll get much better. At the same time, it is likely that many companies are forcing agents and usage onto people that actually will get very little out of them in their current form, which helps the AI buildout (or bubble) continue.
The best closed labs â right now this list is just Anthropic and OpenAI, but itâs reasonable to expect Google to catch up â will always make the most efficient models for intelligence at a given cost. Building models is a mass capital investment of talent, data, and compute. These systems, a combination of model weights, harnesses, tools, and serving infrastructure have massive returns on integration (where open models are designed to work across many, diverse serving situations). These integration benefits â the integration of hardware and new forms of software â can be expressed in any possible way of making models better.
The models in the near future may saturate on benchmark scores, but if that intelligence ceiling really is a cap on utility then the labs will optimize utility per second or per watt, serving users in another way. Improving the models is possible in every direction â there have been no walls in progress. Weâre early in the mass buildout of intelligence, which involves harnessing the physical world to build numerous datacenters, organizing many AI researchers so that a large team can contribute to one model, and of course solving many small, low-level puzzles that unlock performance. Every indication is that there is still meaningful performance to be unlocked and the closed labs are the best set up to extract it.
The collective wisdom of the labs is that making the models smarter, in terms of the frontier of absolute intelligence, has the most value. This is the right call to me because it unlocks large new markets. Optimizing models at a fixed intelligence level locks in markets, expands accessibility over time, and increases return on investment for users (while potentially lowering margins for selling intelligence).
Many people are making this bet that models will keep getting better and are learning to work well in these harnesses, even though some workflows are still a bit clunky. This is the right bet. These people all will continue to use the absolutely best models available. Itâs like buying an iPhone as a consumer. You could get an Android and suffer from a bunch of paper cuts to save money, but why would you? The returns to performance are even higher in the workplace, which drives pricing power.
In this mental model, the frontier labs as businesses, will look like new, reimagined forms of a mix of Apple and Microsoft. The Apple side is that theyâre selling an integrated, extremely hard to replicate technology. The Microsoft side is selling high-leverage subscriptions across the economy. In 5-10 years I expect both OpenAI and Anthropic to be valued in the $2-10T range. The true frontier labs will be an oligopoly that looks like the cloud market today.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
On the other side of this equation is the open model economy. This isnât to say that the frontier labs will dominate all aspects of AI use. Yes, I expect OpenAI and Anthropic to be the most representative companies of the AI boom (new companies, alongside Nvidia of course), but the collective value capture around open models will be far bigger overall, itâs just that the revenue and margins will be shared across a wide stack of companies.
Many businesses want to switch to open models but the models today are not good enough in out-of-distribution tasks. Eventually open model builders will stop chasing Claude and GPT on the Artificial Analysis index and fill this niche. This fork could be driven by economic factors, where they no longer have the revenue to support the growing R&D costs for continuing to scale models. It can also be driven by pure demand, where certain AI solutions only can exist at low price points present in open models. Where closed labs are an oligopoly, open model builders and users will be far more diverse and numerous. The total market value will dramatically exceed the cumulative value of OpenAI and Anthropic.
Open models are by their nature not integrated, so they will rely on multiple companies coordinating to serve them. Each of these layers will have alternatives, driving prices down to commodity pricing. These low, predictable prices will be where many enterprises enter to build in-house agents and tools for niche tasks. The predominant mode of deployment here is that enterprises find a model that hits a sufficient performance threshold on a task of interest and does not replace the model later (setup costs are high). As customizing models becomes easier, again in the open model finetuning stack we are seeing emerge (Tinker, Fireworks, Prime Intellect, etc.), this market becomes even bigger.
What this will look like in the coming years is a steady rise in open model inference proportion across the entrenched hyper-scale clouds of Google, Amazon, Microsoft and new AI infrastructure companies of Together, Fireworks, OpenRouter, etc when compared to OpenAI and Anthropic.
The key is that the open and closed model economies are operating on different exponentials. I still believe that progress will continue at a fast pace across the entire ecosystem, but claims of recursive self improvement (RSI) giving the closed labs an unassailable advantage are overblown. New forms of products like background agents can support both these open and closed models.
The closed models hit incredible product-market fit with the current agents, starting their integrated exponential by monetizing the top end of the knowledge work. The open model economy will take far longer, but it will also be far more satisfying to follow, as it tracks the broader diffusion of AI into the entire economy and world.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
As the years of AI progress go by, itâs been accompanied by a slowly rising tide of consequence. Models are getting more capable, how we work is changing quickly, economics of AI are becoming real, just as real-world risks come to the forefront. 2026 is the first year where I donât think thereâll be any breaks from this. The hard part to prepare for is that thereâs a good chance things just continue to ratchet up from here â more disruption, more surprises, more stakes.
On my end, thereâs been a growing list of topics that are very fateful to how I see the current state of AI, but I havenât even gotten to write about them (at least not from all the angles I want to)! All of these are closely related to the implications of different models reaching new capability levels and how I use that to infer what may come next.
1. Open models havenât had their true agent moment like Opus 4.5
The time gap between open and closed models is very often discussed, but the reality is that we have a nice time-gating thatâs independent of debatable benchmarks â if open-weight models do or do not become super useful in agentic harnesses. The Opus 4.5 in Claude Code moment of December 2025 was so loud and obvious, that if open models hit this performance level for price points as low as $5/month, there will be an explosion in usage.
Right now we are about 5-6 months in with no equivalent open model. I suspect the robustness of the best closed frontier models that I write about could make this moment take a good amount longer, say closer to 12+ months. In this time, Claude Code and Codex may seem like different categories of products. In the standard flurry of new, state-of-the-art open models from a variety of labs, benchmarks will definitely keep climbing, but the open-closed gap should become more interpretable as real-world use becomes the real litmus test.
2. Gemini still doesnât have a meaningful competitor for Claude Code and Codex
The best exclamation point I can offer to reinforce my prediction that open models are further behind than the benchmarks claim is that even the mighty Google doesnât have a clear competitor for Claude Code and Codex. Iâm sure the Gemini team is pushing very hard on this.
I still need to do a lot more testing on Gemini 3.5 Flash, but reading reviews makes it clear that itâs not a substitute for how Iâm working today. Itâs maybe not the Gemini team explicitly specializing for Googleâs existing products (search, YouTube, etc.), but the model seems to suit them. If Google doesnât have a powerful tool here soon, I donât expect the open model labs to either. The open models are going to be used more for automated, enterprise agents and low-cost domains, rather than being the driving tool of modern knowledge work. This will feed directly into the economic engine of funding future models, where the agents like Claude Code and Codex are the current best path to massive AI revenue growth.
I discussed how the current environment is quietly driving labs in China to specialize on AI Proem with Grace Shao and this is central to my expectations of open models specializing over the next few years instead of competing with OpenAI, Anthropic, and Google.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
3. I donât expect an open-weights Mythos this year
While I donât think Mythos is a general âgod modelâ that will crush the competition in every domain, I do think itâs a remarkable technical achievement in software engineering and cybersecurity. Mythos is obviously a watershed moment for those fields. Having spoken to most of the Chinese labs â particularly those with the most prominent, large, open MoE models like Kimi, Z.ai, DeepSeek, and Qwen â I think theyâre heavily resource limited and donât have an immediate path to scaling up training processes like the big labs in the U.S. For the labs which are more corporate, which comes with more resources, such as Alibaba and Bytedance, they also have more conservative stances on safety and security.Mythos is a bellwether of the massive acceleration in training and research compute available to the largest American companies.
Epoch AI recently had a nice piece on the compute available to various labs (~Google 25%, Meta 11%, OpenAI 11%, Anthropic 6%). All of these numbers are vastly higher than any Chinese lab.
4. American open models are slowly gaining steam
Nvidia with Nemotron, Google with Gemma, Arcee AI and others are slowly stabilizing the open model ecosystem in the U.S. Thereâs a lot thatâs hard to measure here, especially in the rise of local agents like OpenClaw and Hermes, but there are adoption numbers of American models that we havenât seen since Llama 3.Gemma 4âs models are all tying or outperforming the equivalently sized Qwen 3.5/3.6 models â where Qwen has for years now been the default open model at these sizes. These Qwen 3.5/3.6 models have been tricky to get working in a lot of post-training research, partially due to architecture/tooling and partially likely due to modeling (i.e. the model is not easy to finetune for some training decision). Iâve heard few complaints about Gemma, but it also could be because Gemma is not yet the researcher default.
There's a simple reality that we've seen recently with models like GPT-OSS, Nemotron 3, and now Gemma 4, that if a model is in the right range of benchmarks and released by an American lab with a truly permissive license, it'll get a large amount of adoption (in this cycle, recall that Gemma 4 adopted the Apache 2.0 License, changing from one with use-case restrictions on earlier Gemmas). This early phase of American growth in open models is establishing key brands directly with developers. The consensus is that more neolabs like Reflection and Thinking Machines are likely to participate in this space, but being too patient will lose the time when new agentic workflows and enterprise relationships are built.
5. Anthropic and OpenAI are just getting up to speed in model iterations
I expect the rest of this year to be a ruthless competition between these two flagship companies. Iâm at an interesting balance where I think GPT 5.5 is a bit smarter of a model and I love the Codex App, so Iâm structuring much of my work to be possible there. At the same time, for a lot of writing-related and broader surface area tasks I really still love Claude. These models are rapidly changing how we work, I run Codex from my phone while doing other things, am setting up automated open model analysis jobs on the back of agents, and expect to be able to scale the research side of Interconnects widely.
AI is beginning to drive companies to the two extremes in the scaling era. The biggest companies will be way bigger than ever, using resources and mass talent to have sustained progress at the frontier of raw AI capabilities. On the other side, tiny businesses like Interconnects thrive by using agents to refine, present, and sell niche expertise. The mass social job displacement thatâll come is going to reduce employability for various knowledge workers that donât fit into either of these extremes for the raw technical side (big or small companies), while sustaining and maybe even amplifying careers that interface directly with humans (e.g. doctors) or other power structures with means to sustain themselves (law/government).
6. More existing power structures will assert themselves on AI
Just in the last few days while writing this, we had the Pope release an over 40,000 word document on where AI is going and China expand personnel movement restrictions on top AI researchers across industry. At the same time, the U.S. has designated Anthropic a supply chain risk and continues to use its models for national security. The list of news like this is only going to grow. Existing power structures are realizing thereâs a finite time window for them to exert themselves in the AI dynamic â an intuition that could be mapped to influence going down as AI models get more powerful. This intuition is potentially dangerous, as it sets up meaningful conflict in who controls the technology (as I discussed with Dean Ball after the Anthropic-DoW spat).
Next: Where technical becomes social
These largely technical and power trends accelerating are going to put more pressure on the social and political anti-AI sentiments within the U.S. This is currently the most obvious barrier to continued AI development and beneficial diffusion. Reflecting on this, many people in the tech discourse get too focused on the details, where yes a lot of data-center-detractors are making genuinely wrong factual claims in defense of their position.
The real position that a large swath of Americans has is that they have a voice in saying no to the current trend â by not granting permission to build data centers. This is a voice that they havenât been granted by the tech industry that changed the face of the global economy and power structures in the last few decades.
This is setting us up for a challenging year ahead for the industry. The labs are aggregating and concentrating talent to peak levels. There are few neutral messengers to communicate the reality of AI to the public. The frontier labs leadership is largely gearing up to IPO and stay ahead in the capabilities race. With the status quo, there are few actions to unwind this path toward social conflict.
It takes individuals in the AI ecosystem to zag and go against the groupthink of needing to make your wealth today, of needing to be at a lab to do impactful work, and so on. Iâm personally continuing to bet on this, by trying to make a vibrant and diverse open model ecosystem supported by clear, unbiased information. If you agree with this and have been watching from the sidelines, itâs a good time to get involved, before the situation spirals into something uncontrollable.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Staring out the window on a new, high-speed train from Hangzhou to Shanghai Iâm gifted with views of dramatic ridgelines speckled with wind turbines that are silhouetted against the setting sun. The mountains cast a backdrop to a mix of spanning fields and clustered skyscrapers. Iâm returning from China with great humility. Itâs a very warming, human experience to go somewhere so foreign and be so welcomed. I had the honor of meeting so many people in the AI ecosystem who I knew from afar, and they greeted me with big smiles and cheer, reminding me how global my work and the AI ecosystem is.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
The mentality of Chinese researchers
The Chinese companies building language models are set up as the perfect fast-followers for the technology, building on long-standing cultural traditions in education and work, along with subtly different approaches to building technology companies. When you look at the outputs, the latest, biggest models enabling agentic workflows, and the ingredients, excellent scientists, large-scale data, and accelerated computing, the Chinese and American labs look largely similar. The lasting differences emerge in how these are organized and conditioned.
Iâve long thought that a reason that the Chinese labs are so good at catching up and keeping up with the frontier is that theyâre culturally aligned for this task, but without talking to people directly I felt like it wasnât my place to attribute substantial influence to this hunch. Speaking with many wonderful, humble, and open scientists at the leading Chinese labs has crystallized a lot of my beliefs.
So much of building the best LLMs today comes down to meticulous work across the entire stack, from data to architecture details and RL algorithm implementations. All points of the model can give some improvements, and fitting them in together is a complex process where the work of some brilliant individuals needs to get shelved in favor of the overall model maximizing a multi-objective optimization.
Where American researchers are obviously also brilliant at solving the individual components, thereâs more of a culture of speaking up for yourself in the U.S. As a scientist, youâre more successful when you speak up for your work and modern culture is pushing the new path to fame of âleading AI scientistsâ. This results in direct conflict. The Llama organization is heavily rumored to have collapsed under the political weight of these interests embedding themselves in a hierarchical organization. Iâve heard of other labs saying that it can be needed to pay off a top researcher to get them to stop complaining about their idea not making it in the final model. Whether or not thatâs exactly true, the idea is clear. Ego and desires for career advancement do get in the way of making the best models. A small, directional shift in this sort of culture between the U.S. and China can have a meaningful impact on the final outputs.
Some of this has to do with who is building the models in China. Thereâs an immediate reality at all of the labs that a large proportion of the core contributors are active students. The labs are quite young, and it reminds me of our setup at Ai2, where students are seen as peers and directly integrated in the LLM team. This is incredibly different from the top labs in the US, where the likes of OpenAI, Anthropic, Cursor, etc. simply donât offer internships. Other companies like Google nominally have internships related to Gemini, but thereâs a lot of concern about whether your internship will be siloed and away from anything real.
To summarize how the slight change in culture can improve the ability to build models:
* More willingness to do non-flashy work in order to improve the final model,
* People new to building AI can be free of prior phases of AI hype cycles, allowing them to adapt to the new modern techniques faster (in fact, one of the Chinese scientists I talked to really actively attached to this strength),
* Less ego enabling org charts to scale slightly, as thereâs less gamifying the system, and
* Abundant talent well-suited to solving problems with a proof of concept elsewhere, etc.
This slight inclination towards skills that complement building todayâs language models stands in contrast to a known stereotype that Chinese researchers tend to produce less creative, field-spawning, 0-to-1 academic style research. Among the more academic lab visits on our trip, many leaders talk about cultivating this more ambitious research culture. At the same time, some technical leaders we talked to were skeptical about whether such a rewiring in the approach to science is likely in the near term, because itâll take a redesign of the education and incentive systems that is too big to happen within the current economic equilibrium. This culture seems to be training students and engineers that are excellent at the LLM building game. They also, of course, have an extremely abundant quantity.
These students told me about a similar brain drain happening in China as in the U.S., where many who previously considered academic paths now intend to stay in industry. The funniest quote was from a researcher who was interested in being a professor to be close to the education system, but remarked that education is solved with LLMs â âwhy would a student talk to me!â
The students have a benefit of coming at LLMs with fresh eyes. Over the last few years weâve seen the key paradigm of LLMs shift from scaling MoEâs, to scaling RL, to enabling agents. Doing any of these well involves absorbing an insane amount of context quickly, both from the broader literature and the technical stack at your company. Students are used to doing this and excited to humbly drop all presumptions about what should work. They dive in head first and dedicate their life to getting the chance to improve the models.
These students are also so magically direct and free of some of the philosophical chatter that can distract scientists. When asking questions on how they feel about the economics or long-term social risks of models, far fewer Chinese researchers have sophisticated opinions and a drive to influence this. Their role is to build the best model.
This difference is subtle, and easy to deny, but it is best felt when having long conversations with an elegant, brilliant researcher who can clearly communicate well in English, basic questions on more philosophical aspects of AI hang in the air with a simple confusion. Itâs a category error to them. One researcher even quoted the famous Dan Wang premise of China being run by engineers, relative to the lawyers of the U.S. when probing in these areas, to emphasize their desire to build. Thereâs no track in China that systematically enables the growth of star power for Chinese scientists, akin to mega mainstream podcasts like Dwarkesh or Lex.
Trying to get Chinese scientists to comment on the coming economic uncertainty fueled by AI, questions beyond the capabilities of simple AGI, or moral debates on how models should behave all served to capture the upbringing and education of these scientists (edited). They are extremely dedicated to their work, but have grown up in a system where debates and opinions on how society should be structured and changed are not encouraged.
Zooming out â Beijing especially felt much like the Bay Area, where a competitive lab is a short walk or Uber away. I got off a flight and stopped by Alibabaâs Beijing campus on the way to the hotel. Then, in 36 hours we went to all of Z.ai, Moonshot AI, Tsinghua University, Meituan, Xiaomi, and 01.ai. Travel by Didi is easy, and if you select an XL in China youâre often paired with electric mini vans that have massage chairs. We asked the researchers about the talent wars, and they said itâs very similar to what weâre experiencing in the U.S. Itâs normal for researchers to bounce around, and much of where people choose to go is based on the best current vibes.
In China, the LLM community feels far more like an ecosystem than battling tribes. Across many off the record conversations, itâs nothing but respect for peers. All of the Chinese labs fear Bytedance with their popular Doubao model, which is the only frontier closed lab in China. At the same time, all of the labs have massive respect for DeepSeek as the lab with the best research taste in execution. When you meet with lab members off the record in the States, sparks fly quickly.
The most striking part of the humility of Chinese researchers is how they also often shrug on the business side, saying itâs not their problem, where everyone in the U.S. seems to be obsessed with various ecosystem-level industrial trends, from data sellers to compute or fundraising.
Where Chinaâs AI industry differs (and matches) the Western labs
The thing that makes building an AI model today so interesting is that itâs not just about getting a group of great researchers in one building together to produce an engineering marvel. It used to be this, but to sustain AI businesses, the LLMs are becoming a mix of building, deploying, funding, and getting adoption for this creation. The leading AI companies exist in complex ecosystems that supply money, compute, data and more in order to keep pushing the frontier.
The integration of these various inputs to creating and sustaining LLMs is fairly well conceptualized and mapped for the Western ecosystem, as typified by Anthropic and OpenAI, so finding big differences in how the Chinese labs think about it points at where the different companies can be making meaningfully different bets on the future. Of course, these futures can be heavily dictated by the constraints on funding and/or compute.
Iâve documented the biggest âAI Industryâ level take-aways from talking to these labs:
* Early signs of domestic AI demand. Thereâs a much-touted hypothesis that the Chinese AI market will be smaller because Chinese companies donât tend to pay for software â thus, never unlocking a giant inference market supporting labs. This is only true for software spend that maps to the SaaS ecosystem, which is historically tiny in China, where on the other hand there is obviously still a large cloud market in China. A crucial unanswered question â one which the Chinese labs themselves debate â on if spending for AI in the enterprise tracks the SaaS market (small) or the cloud market (fundamental). On net, it feels like AI is trending closer to the cloud, and no one was actively worried about a market growing around the new tools.
* Most developers are Claude-pilled. Most of the AI developers in China are obsessed with Claude and how itâs changed how they build software, despite Claude nominally being banned in China. Just because China has historically been hesitant to buy software does not give me the impression that there wonât be a massive surge in inference demand. Chinese technical staff are so practical, humble, and motivated â a fact that seems stronger than any commitment to previous habits in not spending.Some Chinese researchers mention building with their own tools, such as the Kimi or GLM CLIs, but all of them mention building with Claude. There were also surprisingly few mentions of Codex, which is definitely surging in popularity in the Bay Area.
* Chinese companies have a technology ownership mentality. The Chinese culture is combining with a roaring economic engine to create unpredictable outcomes. Iâm left with a lasting feeling that the numerous AI models reflect a practical, current equilibrium of the many technology businesses here. Thereâs no master plan. The industry is defined by a respect for ByteDance and Alibaba, the incumbents expected to win large portions of all markets with their substantial resources. DeepSeek is the respected technical leader, but far from a market leader. They set the direction, but arenât set up to win economically.This leaves companies like Meituan or Ant Group, where people in the West can be surprised theyâre building these models. In reality, they see LLMs obviously as being central to future technology products, so they need a strong base. When they fine-tune the strong, general purpose model it hardens their stack from getting the open community to provide feedback on it, and they can keep internal, fine-tuned versions of the model for their products. The âopen-firstâ mentality in the industry is largely defined by practicality â it helps make their models get strong feedback, it gives back to the open-source community, and empowers their mission.
* Government aid is real, but unclear how big. Itâs often asserted that the Chinese government is actively helping with the open LLM race. This is a government thatâs decentralized across many levels, each of which doesnât have a clear playbook for what exactly they do. Neighborhoods in Beijing compete for tech companies to house their offices there. The âhelpâ offered to these companies almost certainly involved removing bureaucratic red tape like permits, but how far does it go? Can levels of the government help attract talent? Can they help smuggle chips? Across the visit, there were many mentions of government interest or help, but far too little to report the details as assertive or have a confident worldview of how government can bend the trajectory of AI in China. There were certainly no hints of the top levels of the Chinese government influencing any technical decisions in the models.
* The data industry is far less developed. Having heard so much about the likes of Anthropic or OpenAI spending $10M+ for single environments, with cumulative spend on the order of hundreds of millions per year to push the frontier of RL, we were eager to know if Chinese labs are either buying the same environments from companies in the U.S. or supported by a mirrored domestic ecosystem. The answer was not quite complete that thereâs no data industry, but rather that their experience was that the data industry was relatively poor quality and it is often better to build the environments or data in-house. Researchers themselves spend meaningful time making the RL training environments, and some of the bigger companies like ByteDance and Alibaba can have in-house data labelling teams to support this. This all mirrors the build-not-buy mentality from the previous bullet.
* Desperation for more Nvidia chips. Nvidia compute is the gold-standard for training and everyone is limited in progress by not having more of it. If supply was there, it is obvious that they would buy it. Other accelerators, including but not limited to Huawei, were spoken positively of for inference. Countless labs have access to Huawei chips.
These points paint a very different picture of an AI ecosystem, where quickly mapping how Western labs operate to their Chinese counterparts will often result in a category error. The crucial question is if these different ecosystems will produce meaningfully different types of models, or if the Chinese models will always be explained by being similar to the U.S. frontier models of 3-9 months ago.
Conclusion: The global equilibrium
I knew so little about China going into the trip and came out with the feeling of just starting to learn. China isnât a place that can be expressed by rules or recipes, but one with very different dynamics and chemistry. The culture is so old, so deep, and still completely intertwined with how domestic technology is built. I have much more learning ahead.
So much of the current power structures in the US use their current worldviews of China as crucial mental devices for decision making. Having talked, in person, either formally or informally to pretty much every leading AI lab in China, there are a lot of qualities and instincts in China thatâll be very hard to model with Western decision making. Even after asking directly about why these labs release their top models openly, the intersection between ownership mentality and genuine ecosystem support is hard for me to connect the dots on.
The labs here are practical and not necessarily absolutists around open-source, where every model they build would be released openly, but thereâs a deep intentionality in supporting developers, the ecosystem, and using it as a way to learn more about their models.
Almost every major Chinese technology company is building their own general purpose LLMs, as we see with the likes of Meituan (delivery service) and Xiaomi (broad consumer technology company) releasing open weight models. The equivalent companies in the U.S. would just buy services. These companies arenât building LLMs out of a race to be relevant with the hot new thing, but a deep fundamental yearning to control their own stack and develop the most important technologies of the day. When I look up from my laptop and always see bunches of cranes on the horizon, it obviously fits in the with the broader culture and energy around building in China.
The humanity, charm, and genuine warmth of Chinese researchers is extremely humanizing. At a personal level, the cut-throat geopolitical conversation weâre used to in the U.S. hasnât permeated them at all. The world can use more of this simple positivity. As a citizen of the AI community, I currently worry more about the fissures appearing within members and groups around labels of nationality.
Iâd be lying if I said I didnât want US labs to be clear leaders in every part of the AI stack â especially with open models where I spend my time â Iâm American, and thatâs an honest preference. With this, I want the open ecosystem itself to thrive globally, as this can create safer, more accessible, and more useful AI for the world, and right now the question is whether American labs will take the steps to own that leadership position.
As of finishing this piece, more rumors are swirling of executive orders influencing open models, which can further complicate this synergy between American leadership and the global ecosystem â it doesnât fill me with confidence.
Thank you to all the wonderful people I got to talk to at Moonshot, Zhipu, Meituan, Xiaomi, Qwen, Ant Ling, 01.ai, and others. Everyone has been so welcoming and gracious with their time. Iâll keep sharing my thoughts on China as they crystallize, across culture generally and AI specifically. It is obvious that this knowledge will be directly relevant to the story unfolding at the frontier of AI development.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
âDistillation attacksâ is a horrible term for what is happening right now. Yes, some Chinese labs are hacking or jailbreaking APIs to attempt to extract more signal from model APIs â stopping this is important to maintain the U.S.âs lead in AI capabilities. Referring to this as distillation attack is going to irrevocably associate all distillation with this behavior, and distillation generally is a core technique needed to diffuse AI capabilities broadly through academic and economic activities.
We went through this sort of language transition with the open source vs open weight debate. All the terms just reduced to open models â very few people in the large AI community know exactly how open-source differs from open-weights. And terminology matters, as the less informed people who still care about â and influence â the technology are bound by different terms they use. If weâre not careful with the discourse around distillation, many people could associate this broad technique used for research and development of new models as an act at the boundary of corporate manipulation and crime.
Iâve recently written a more technical piece on estimating how impactful state-of-the-art distillation methods are on leading Chinese models, and this piece follows to push for caution in any hasty actions to target the methods with policy. To set the stage, recall Anthropicâs recent blog post where they detailed âdistillation attacksâ made by 3 Chinese labs.
These labs used a technique called âdistillation,â which involves training a less capable model on the outputs of a stronger one. Distillation is a widely used and legitimate training method. For example, frontier AI labs routinely distill their own models to create smaller, cheaper versions for their customers. But distillation can also be used for illicit purposes: competitors can use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently.
This is a clever paragraph, where they normalize distillation generally and explain how a few people can use it illicitly, without detailing how illicit use often involves other more explicit behavior like jailbreaking, hacking, or identity spoofing of the API.
Distillation itself is an industry standard. Itâs used extensively, primarily in post-training, by smaller players to create specialized or smaller models. In my book coming this summer, I describe it as follows:
The term distillation has been the most powerful form of discussion around the role of synthetic data in language models. Distillation as a term comes from a technical definition of teacher-student knowledge distillation from the deep learning literature.
Distillation colloquially refers to using the outputs from a stronger model to train a smaller model.
In post-training, this general notion of distillation takes two common forms:
* As a data engine to use across wide swaths of the post-training process: Completions for instructions, preference data (or Constitutional AI), or verification for RL.
* To transfer specific skills from a stronger model to a weaker model, which is often done for specific skills such as mathematical reasoning or coding.
With this definition, itâs easy to see how distillation takes many forms. Of course, if you just take the outputs from GPT-5.5 and train a recent open-weight base model with them to host a competitive product, thatâs one thing. But, a lot of the things that fall under the bucket of distillation are complex, multi-stage processes that muddle the exact impact of the model you distilled from.
Modern LLM processes could look like using a GPT API to build an initial batch of synthetic data to build a specialized small data-processing model. A good example is a model like olmOCR (or many other models in this category) that are trained to convert PDFs to clean text. This specialized model would be used to create large amounts of data. Finally, you train another model (often from scratch) with the new data you created. Is this final model distilled from GPT?
When done via a closed, API-based model, distillation sits in the grey area of the terms of service that you agree to when signing up to the Claude or GPT platform. They generally forbid the use of the API to create competing language model products, but this term has largely gone unenforced. The open-source community used to worry deeply at being cut off from these cutting-edge APIs for doing research or creating public datasets, but to date only one prominent case of corporate accounts being restricted exists (at least until the recent Chinese companies).
This is all to say that distillation is an industry standard technique, and the use of closed APIs to perform distillation has always been a grey area. Nvidiaâs latest Nemotron models, as one of the only models with open post-training datasets, are technically in large part distilled from Chinese, open-weight models. The Olmo models weâve built at Ai2 are distilled from a mix of open and closed models. This grey area was brought to the forefront again when it turned out that xAI has been distilling from OpenAI. Quoting from the recent trial proceedings between Elon and OpenAI:
OpenAIâs counsel asked Musk whether xAI has ever âdistilledâ technology from OpenAI.
Musk: âGenerally AI companies distill other AI companies.â
âIs that a yes?â Savitt asked.
Musk: âPartly.â
xAI is likely the largest, and most successful AI company willing to thread the grey area that is distillation from their competitors. On the other side, the majority of startups and research groups with fewer resources than them have very likely engaged in distillation of some capacity from Claude, GPT, or Gemini models.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
In the above Anthropic blog post, the problem with the distillation attacks by a few Chinese labs is less the distillation and more the means of attack. It is documented that Chinese labs are actively working to get around the intended use of the API, e.g. to provide additional reasoning data that is very useful for training.
Of course no one should be able to access information from a model that a developer didnât intend to reveal in their APIs (e.g., reasoning traces which would be helpful for training). Associating all of distillation with these attacks, which is to date an industry standard for post-training, from open and closed models alike will be a massive own goal.
What these few labs are doing should be referred to as jailbreaking or abuse, rather than distillation.
The discourse around these actions is creating a troubling discussion thatâs marching towards a mix of regulatory capture or regulatory exuberance thatâs most likely to harm the U.S.âs ecosystem more than Chinaâs. Even if we ban, most likely through potential legal action and other penalties, this type of API abuse, the Chinese companies will likely still do it. Weâve seen this playbook with Chinese multimedia models taking a flexible view of copyrighted content that no U.S. player is willing to take the risk on.
This distillation discussion has quickly snowballed, with a bill moving out of a committee in Congress, an executive order pushing for action, and congressional oversight targeting U.S. companies building on Chinese models (which are downstream of distillation). This multi-pronged regulatory environment could yield truly horrible outcomes â such as figuring out a way to effectively ban open-weight models in the U.S. that are built in China by groups abusing closed LLM APIs.
It is obvious that no bill will literally ban open models, but they can create grey area that exposes entities to unwanted risk or require certain provisions that are bureaucratically very challenging to fulfill, squashing small open source contributors.
In that scenario, the groups who lose are Western academics and smaller companies building models for the long-tail of AI uses. The ecosystem here could be made permanently irrelevant with the removal of nearly all Chinese open-weight models. There is no immediate substitute and building new models with meaningful community adoption has a lead time measured in 6+ months. In the time it takes to build a new domestic open-source ecosystem, countless researchers wouldâve moved onto closed training platforms or into new areas.
Altogether, Iâm hoping this flurry of discussion around distillation becomes a nothing-burger and not a hasty, multi-pronged policy push. We need to avoid two things:
* A wholesale negative connotation of the word distillation, which is used extensively across the AI ecosystem.
* A domestic ban of the open-weight models built by organizations engaged in some portion of distillation.
In addition to this, I want the leading U.S. AI companies to be able to provide their APIs without having their IP leak. They should share more information on why it is hard for them to secure their APIs, but thatâs an issue out of scope for my expertise.
Iâll conclude with a proposal from my friend Kevin Xu at Interconnected Capital (and great Substack) on why this current distillation dynamic may actually be good for the leading labs.
If all the Chinese companies are addicted to distillation as a way of getting close to the frontier, then theyâll never actually learn the techniques needed to take an outright lead. If we cut off the Chineseâs obvious crutch in model building, weâll gain a short-term lead in AI, but in the long-term that may be what they needed to get on a more competitive long-term trajectory.
This is the same debate weâre having with other technologies where the U.S. currently has a lead, e.g. with advanced semiconductor technologies. So I understand the trade-offs, but we not should crack down on all of distillation.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Weâre living through the period of time when weâll learn if open models can keep up with closed labs. The obvious answer is that no, they wonât. This answer is a form of saying they wonât keep up in every area. This framing closes off a popular prediction where the open models completely catch up, as in all models saturate and open and closed models only become increasingly similar. In living through this, itâs evidently very unclear when the longer-term stable balance of capabilities will solidify.
This is a very complex dynamic, where the core point we monitor is a capability gap between models. At the same time, this gap is intertwined with evolving dynamics in the funding of open models, who builds open models, how techniques like distillation that enable fast-following translate through new application domains, potential regulation hampering the open-source AI ecosystem, and of course who actually uses open models.
The capabilities gap is one signal in a complex sea of forces, pushing supply and demand into different shapes. In many cases the demand â where obviously tons of individuals, organizations, and sovereigns want, or need, open models â is largely separated from supply. Supply is fully dictated by economics. The question of âwhich business strategies support releasing open modelsâ is still at stake.
Interconnects AI is a reader-supported publication. To receive new posts and support my work, consider becoming a subscriber.
With this complexity, I wanted to distill my key beliefs down into a clear list. These are downstream of 10+ pieces Iâve written or recorded on open models this spring (which are linked throughout).
* Itâs surprising that the top closed models did not show a growing capability margin over open models, based on compute differences for training and research, especially in the second half of 2025 and through today.
* Open model labs are technically very strong at keeping pace on well-established benchmarks. This will continue and reflects a balance of abundant talent and sufficient computing power.
* Chinese open-weight labs focus slightly more on benchmark scores than comparable closed labs in the U.S. Distillation helps the Chinese LLM companies do so, but itâs not a panacea. Changes in the distillation dynamic (e.g. regulation) will not be a determining factor on the balance of capabilities. This increase in focus is a natural evolution of their incentives in keeping the narrative on keeping up with the frontier alive, which is crucial to fundraising and adoption.
* To date, closed models tend to be more robust and generally useful than similarly scoring open models. Closed models have certain hard-to-measure qualities that are not well captured in current or past benchmarks. This will be key to enabling closed models to dominate in markets where an individual user constantly presents new challenges, i.e. supporting knowledge workers as a direct assistant.
* The open vs. closed model race, as monitored through benchmarks, will largely be a game of economic staying power and fast-following, until the market structure constricts. I expect Chinese open-weight labs to face funding difficulties first, as soon as later this year. Funding difficulties will be seen in different capability trajectories 3-9 months later.
* The RL dominated training era has increased the relevance of distribution to real-world use-cases as a key factor in continued capabilities improvements. These are tasks where users directly use tools like Claude Code or Codex to solve problems in their job with agents. This is the first clear technical area that closed labs can dominate open-weight models on capabilities, potentially leveraging online RL directly based on user feedback.
* Open models will be increasingly adopted in repetitive automation tasks, as measured in the relative share of the API market, for repetitive tasks across the ecosystem. This takes the form of many new AI-native applications, business backend automation, etc. The success of this will drive more investment in domain-specific, efficient open models.
This is a complex picture, where the long-term trajectory is more of an economics question rather than an ability one. Many other outlets can paint a far more simplistic narrative that âChina will assuredly catch us in AIâ and get more distribution because it is a simple story. The reality is complex. Only real AI revenue begets more investment, eventually thatâll be linked to the ability to keep improving models at a rapid rate. Economic realities have not yet impacted scaling open models, as a general category.
This economic-focused angle relates to my positions on the open model ecosystem more broadly.
* Recurring calls to ban certain types of open models will continue to come but are in practice impossible to implement. Training strong AI models (i.e. near but not at the frontier) is a relatively small cost compared to large-scale deployments. E.g. if the U.S. bans open models over a certain compute threshold, another sovereign entity will eventually train them and release them publicly, with the models entering the U.S. market with less oversight.
* The second derivative of influence on open models has shifted, and the U.S. will slowly regain ground in adoption metrics of open models starting in early 2027 (it takes a long time for Chinaâs velocity to slow, then flip). Examples include Googleâs Gemma 4 (a wild success), Nvidiaâs Nemotron, and Arcee AI.
* As ever-stronger closed models are built, previewed, and released, there will be more safety-shocks saying that open-weight versions of the strongest AI models never can be allowed to exist, similar to reactions to Claude Mythos. These can spur burdensome regulation on open models.
* With the above, there will also be increased long-term interest in open models, as sovereign entities and existing power structures realize the coming, super powerful AI tools cannot land in the hands of only one or a few companies. These entities will see open models as a different governance paradigm.
* New funding structures for open models will emerge, as many stakeholders realize dependencies on single, for-profit companies for access to intelligence are unreliable.
* Local agents, OpenClaw, and other personal agents represent a large, to date, mostly ignored market for open model usage. It is a sort of dark matter, with pervasive, massive potential for influence on the balance of open-to-closed models.
A single word governs this post and is intentionally repeated â complex.
This complex reality has been driving me to think more deeply about how to clearly describe the open model gap, and why I can hold it in my head that I expect American closed labs to clearly draw ahead, despite the fairly unequivocal evidence in support of the capabilities of recent open-weight models. More on the nuance in the open-closed gap in another piece coming soon, so please subscribe!
Let me know any positions that I missed.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Recently, I was talking with Percy Liang, Stanford professor and lead of the Marin project (another fully-open model lab), and it set in on me that there will eventually be a consortium of companies funding a foundational set of open models used across industry. Itâs not clear when thisâll emerge, and Nemotron (Coalition) is Nvidiaâs attempt to bankroll and bootstrap this approach within a single wealthy company, but a consortium is the only long-term stable path to well-funded, near-frontier open models.
In recent months, weâve seen a lot of turnover in open model labs, with high-profile departures at Qwen and Ai2 (my comment). This shouldnât be super surprising to followers of the ecosystem â itâs happened before with Meta shifting its focus away from Llama, and itâll only happen more as the cost of trying to keep pace at the frontier of AI only increases. The other leading labs with models available today include Chinese startups such as Moonshot AI, MiniMax, and Z.ai â all of which look precarious on their ability to fund continued growth in the cost of training or R&D. Releasing oneâs strongest models openly today is in active tension with the option of spending focus and resources on AI products that can currently generate meaningful revenue (and profits).
Weâre going to see business models emerge around releasing some, or even many, models openly, but these will largely be smaller models that enable a long-tail of functionality, rather than models at the absolute frontier. This class of companies thatâll release many, strong fine-tunable models will include the likes of Arcee AI, Thinking Machines, OpenAI, Google with Gemma, and more in that class. The cost and relative advantage of keeping the best models closed in a business environment with many opportunities for revenue are too high. To summarize â there will be an ever increasing number of companies releasing models that are good for creating a lively niche of smaller, custom models, but an ever decreasing number of companies willing to release fully open, near-frontier models.
This is the core thesis of why Iâm pushing hard for more people to do more research on how these smaller models can complement the best closed agents, the science of finetunability, etc. See my post below â itâs about creating a sustainable open model ecosystem, whether or not the frontier of open keeps paced with closed:
Itâll take years for this equilibrium to become more obvious, seen through the lens of more open model families coming and going. This year, it seems likely weâll see Nvidiaâs Nemotron reach new heights, Reflection AI challenge some of the Chinese models with a strong, large MoE, maybe Meta releases a new open-weight model, and so on. True pressure to change strategy will only come when the capital environment punishes the less efficient spend on resources (e.g. giving away your competitive advantage, in having an in-house model). This pressure will likely hit Chinese startups training these models first.
All of Moonshot AI, MiniMax, and Zhipu AI will show signs of financial challenge in the coming years if they retain their strategy, on top of their models falling further behind the best open models in terms of generality. This is inevitable pressure to evolve open models to areas that are profitable and complementary of the frontier of AI.
Nvidia, which is best positioned to support the open ecosystem in the near term to support its core GPU business, could face many pressures to pull back its open model efforts. It could:
* Realize itâs too competitive to their biggest customers as they succeed too much with Nemotron,
* Fall to competition on their core business and lose the free cash flow buffer needed to fund this (e.g. itâs 2031 and OpenAI, Anthropic, Google, and the other frontier labs are worth so much they build their own chips).
* Start succeeding beyond their initial goals and keep the chips for them to build ASI themselves, as a closed-weight model.
The pressures for new funding mechanisms for open models are based on the assumptions of continued, substantive progress on the capabilities of frontier models. Mechanisms such as self-improvement and scaling all stages of the training pipeline are underway. This progress of capabilities will only increase the potential profit in selling models as and in products, not giving them away. The scale of investment required has already begun to push away non-profits from the game of making truly frontier-scale models. Capitalism is designed to make companies ruthless and chase down leads on profitability, not donate technology as charity.
As the economic environment shifts companies away from releasing the strongest models openly, more companies that rely on these models will look for an outlet of securing model access into the future. This is going to be compounded by a growing group of companies who come to rely on open-weight models for their workflows.
These points loop back into how model training is getting more expensive, so where desire to have the models will go up, ability to procure them will go down for many players. There are x-factors that could multiply the demand for institutions to ensure the existence of open models, such as the best frontier models not even being available via API (such as if Claude Mythos never goes general access).
As training relevant models is shifting to cost billions of dollars, rather than millions, few companies well be able to afford it. many companies will bite at the cost of paying 1/10th of the cost to train a frontier model, or if the consortium works, 1/50th. The upside for companies will be some mechanism to steer development (e.g. model sizes) or getting early access to develop internal and open-source tooling for the model.
It is in my nature to, by default, say this idea will fail, as training models is inherently a complex and high-focus endeavor, one that requires integration of every part of the stack and focusing specifically on your own vision and needs, rather than trying to serve every possible user. Eventually the need for open intelligence â and economic pressure to build it â will make a model consortium inevitable.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
With the announcement of the Claude Mythos model this week and the admittedly very strong stated abilities, especially in cybersecurity, a new wave of anti open-weight AI model narratives surged. The TL;DR of the argument is that our digital infrastructure will not be ready in time for an open-weight version of this model, which will allow attacks to be conducted by numerous parties.
The backlash against open models in the wake of the Mythos news conflates too many general unknowns into a simple, broad policy recommendation that could actually further weaken cybersecurity readiness.
Weâve been here before â open-weight models were discussed as being extremely dangerous when OpenAI withheld GPT-2 weights in 2019, and when OpenAI released GPT-4 in 2023. Both of these waves came and went. The core mistake that is being made is the composition of two issues: 1) the acceptance of the open-closed model gap being static in time and 2) linking open-weight viability generally to specific issues.
Iâve written at length recently on how I think that the best, frontier-level open weight models are going to fall behind the best closed models in overall capabilities in the near future. Iâve also written about how the open-weight ecosystem needs to adapt to accept this reality. This is one of the times for the AI industry where I will repeat that itâs a total blessing to have the 6-18 month delay from when a certain capability is available within a closed lab to it being reproduced in the open. Itâs a good balance of safety and monitoring the frontier of AI systems while allowing a useful open-source ecosystem to exist and thrive.
The core argument Iâve focused on in the open-closed model time gap has been in general capabilities â i.e. for general purpose, frontier models such as Claude Opus 4.X or GPT Thinking 5.X. The abilities of these closed models to robustly solve and work in diverse situations as agents remains out of scope of the best open-weight models. What the open-weight models have tended to be better at is quickly keeping pace on key benchmarks (which admittedly is helped to some extent, but not necessarily substantially by distillation). This discussion is entirely different, it has to do with if open weight models can keep pace on the specific skills related to cybersecurity, and when we could expect an open version of this model to be available to the world.
The case of a Claude Mythos level open weight model is admittedly more nuanced to me than the previous few anti-open weight narratives the community has experienced. Where GPT-4 was about a more hypothetical risk, especially in areas like bio-risk, the clear and present reality of cyber infrastructure being prone to attack is far more tangible. Still, much of this nuance in the moment comes down to not knowing the full details of what the system can actually do (i.e. Mythos), and the state of the environment it would act in (i.e. our digital infrastructure).
To properly assess this risk, we need to know what it takes to build and deploy a Claude Mythos scale model. This entails three pieces: 1) training and releasing the weights, 2) the harness that gives the model effective tools it knows how to use, and 3) the inference compute and software.
(Below I make some model size & price estimates to show my thinking, these should not be taken as ground truth.)
Current estimates put the size ranges of leading models like Claude Opus 4.6 or GPT 5.4 as being around 3-5T parameters. Currently, the largest open-source models, which have been coming from Chinese labs, are around 1T parameters. Claude Mythosâs preview pricing is 5X Opus, which could come from a simple multiplicative increase in active parameters (with the same serving system design), far higher inference-time scaling, more complex harnesses that make inference less efficient, lower utilization expectations, and so on. The simplest guess is that itâs a mix of all of the above, something like 2X bigger in parameters and much less efficient to serve. Thatâs a huge model, likely something similar to GPT 4.5, but actually post-trained well (GPT 4.5 was ahead of its time, infra-wise).
With size comes the challenge actually training the model, as bigger models always come with new technical problems that must be solved to unlock the capabilities. For the case of cybersecurity, my guess is that most of the capabilities can be learned by training a model to be superhuman on coding. Unlike some capabilities such as knowledge work, medicine, law, etc., coding can be studied and improved substantially with public data like GitHub. Iâm far more optimistic in open-weight models staying fairly close to the frontier in narrow domains of code execution and processing, but I donât understand the full scope of skills needed to be superhuman in cybersecurity understanding. How much expert knowledge and special sauce went into training Claude Mythos? Thatâs a substantial source of my error bars on the impact.
Second, we know nothing about how the model works under the hood. Today, models are complex systems that entail far more than just weights. They require complex tools and infrastructure to run them, of which Claude Code is the one we are most used to. Mythos very likely has its own innovations here.
My estimate for how many GPUs youâd need to serve an 8T parameter, modern MoE is something like O(100) H100 GPUs, which costs something like $10K a day (and this may be very slow in terms of tok/s). Heck, the official marketing copy of the Nvidia GB200 VL72 system is âUnlocking Real-Time Trillion-Parameter Modelsâ on the rack. Does Mythos fit on one rack? The point isnât to rely on my specific estimate as a policy reference, but to repeat that running leading AI systems is very expensive and not something you can just do on a laptop or self-service cloud portals.
There are far fewer actors who can get their hands on these resources, relative to those who can download the model. Of course, there are still many, but itâs important to flesh out all the details of what it would take to proliferate the capabilities of a Mythos-like model. In summary, tools like Mythos will make the best attackers have more powerful tools of the trade, but it wonât be handing a nuke to every teenager connected to the internet.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
Personally, I do acknowledge thereâs a chance that cybersecurity abuse is a red line that makes releasing open-weight text models above a certain capability threshold morally grey. Many people thought this red line would come far earlier, somewhere in between GPT-2 and GPT-4, through the harm axis of mis/disinformation, but that had different bottlenecks. For image generation models, weâre well past the first red line which is enabling non-consensual AI deepfakes with readily available open-weight models. Weâre balancing the reality of these fears having come and gone before with a technology thatâs becoming increasingly capable.
So, my second large source of error bars is âhow bad is it actuallyâ with respect to the state of cybersecurity. How much can humans clean up in the most important software with months of private access to a model like Claude Mythos? What will never get fixed?
For example, if we get open-weight models that are close to the capabilities of Claude Mythos, could those be fine-tuned by organizations to harden the security of their tools?
Currently, itâs too soon to call it as a general reason to stop progress in open models. When Claude Mythos is closed to so few partners, in some ways having strong open models close to the threshold makes assessing the danger easier. Having to rely fully on a single private company to determine the security of essential, international infrastructure is not a tenable equilibrium.
So, in conclusion, I urge people to further study three things:
* How do we measure cybersecurity related capabilities across open and closed models. With this, are open models truly keeping up at a 6-9month lag, or are they only maintaining performance relevance in other areas of coding?
* How do we independently measure the true impact of Claude Mythos and Project Glasswing on existing cybersecurity concerns?
* If it is the case that the models are keeping up and the defensive capabilities of Claude Mythos are weak, how do we better monitor (and if needed, try to regulate) the targeted capabilities of open-weight models in narrow domains?
The goal is to encourage fears about open models remaining very specific. Any general ban on open models in a nation will immediately and likely irrevocably remove that entityâs ability to influence a crucial, and amorphous technology. If we stop building the best open models in the U.S., then another country will do this and become the center of the technology. Thereâs no way to fully kill open models, only influencing, understanding, and steering.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Having written a lot of model release blog posts, thereâs something much harder about reviewing open models when they drop relative to closed models, especially in 2026. In recent years, there were so few open models, so when Llama 3 was released most people were still doing research on Llama 2 and super happy to get an update. When Qwen 3 was released, the Llama 4 fiasco had just gone down, and a whole research community was emerging to study RL on Qwen 2.5 â it was a no brainer to upgrade.
Today, when an open model releases, itâs competing with Qwen 3.5, Kimi K2.5, GLM 5, MiniMax M2.5, GPT-OSS, Arcee Large, Nemotron 3, Olmo 3, and others. The space is populated, but still feels full of hidden opportunity. The potential of open models feels like a dark matter, a potential we know is huge, but few clear recipes and examples for how to unlock it are out there. Agentic AI, OpenClaw, and everything brewing in that space is going to spur mass experimentation in open models to complement the likes of Claude and Codex, not replace them.
Especially with open models, the benchmarks at release are an extremely incomplete story. In some ways this is exciting, as new open models have a much higher variance and ability to surprise, but it also points at some structural reasons that make building businesses and great AI experiences around open models harder than the closed alternatives. When a new Claude Opus or GPT drops, spending a few hours with them in my agentic workflows is genuinely a good vibe test. For open models, putting them through this test is a category error.
Something else to be said about open models in the era of agents is that they get out of the debate of integration, harnesses, and tools and let us see close to the ground on what exactly is the ability of just a model. Of course, we canât test some things like search abilities without some tool, but being able to measure exactly the pace of progress of the model alone is a welcome simplification to a systematically opaque AI space.
The list of factors Iâd use to assess a new open-weight model Iâm considering investing in includes:
* Model performance (and size) â how this model performs on benchmarks I care about and how it compares to other models of a similar size.
* Country of origin â some businesses care deeply about provenance, and if a model was built in China or not.
* Model license â if a model needs legal approval for use, uptake will be slower at mid-sized and large companies.
* Tooling at release â many models release with half-broken, or at least substantially slower, implementations in popular software like vLLM, Transformers, SGLANG, etc due to pushing the envelope of architectures or tools.
* Model fine-tunability â how easy or hard it is to modify the given model to your use-case when you actually try and use it.
The core problem is that some of these are immediately available at release, e.g. general performance, license, origin, etc. but others such as tooling take day(s) to week(s) to stabilize, and others are open research questions â with no group systematically monitoring fine-tunability.
In the early era of open models, the days of Llama 2 or 3 and Qwen pre v3.5, the architectures were fairly simple and the models tended to work out of the box. Some of this was due to the extremely hard work of the Llama, Qwen, Mistral, etc. developer teams. Some is due to the new models being genuinely harder to work with. When it comes to something like Qwen 3.5 or Nemotron 3, with hybrid models (either gated delta net or mamba layers), the tooling is very rough at release. Things you would expect to âjust workâ often donât.
Iâve been following this area closely since we released Olmo Hybrid with a similar architecture, and Qwen 3.5 is just starting to work well in the various open-source tools that need to all play nice together for RL research. Thatâs 1.5 months after the release date! This is just to start really investing more into understanding the behavior of the models. Of course, others started working on these models sooner by investing more engineering resources or relying on partially closed software. The fully open and distributed ecosystem takes a long time to get going on some new models.
All of this is lead-in for the most important question for open models â how easy is it to adapt to specific use-cases? This is a different problem for different model sizes. Large MoE open-weight models may be used by entities like Cursor who need complex capabilities in their domain, e.g. Composer 2 trained on Kimi K2.5. Other applications can be built on much smaller models, such as Chromaâs Context-1 model for agentic search, built on GPT-OSS 20B.
The question of âwhich models are fine-tunableâ is largely background knowledge known by engineers across the industry. There should be a thriving research area here to support the open ecosystem model. The first step is to understand characteristics of different base and post-trained models to understand what they look like. The second step is to tune pretraining recipes for open models so theyâre more flexible.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
For The ATOM Project and other Interconnects endeavors, weâve put in substantial effort to measuring adoption trends in the open ecosystem. Everything takes a long time to unfold after a model is first publicly available â and adaptability is why. What we know for sure now, when Qwen has been going from strength to strength with its releases, is that technical staff across the industry has gotten comfortable working with Qwen models. Countless research methods and datasets were made to work with Qwen. Itâll take patience for any other model family to get to this point â a patience Iâm not sure many open model builders have.
This takes us to Gemma 4, Googleâs latest open models. Gemma 3 was released more than a year ago, in March of 2025, and is a bit underrated. Gemma 4 comes in 4 sizes for now, with a bigger, MoE model of over 100B total parameters rumored but not released yet. The models we have today come in sizes of ~5B dense, 8B dense, 26B total 4B active MoE, and 31B dense.
Iâm most excited that theyâre finally adopting a standard Apache 2.0 open source license. Thisâll massively boost adoption. The standard of better licenses for strong open-weight LLMs was set by mostly Chinese open model labs in the last 1-2 years, and now U.S. companies are following suit. I will personally be so happy if the horrible Llama licenses and Gemma terms of service were an ~18-month transient dynamic of the industry being nervous about releasing strong open models.
The Gemma 4 scores look very solid, the small models have incredible benchmark scores (especially in general domains like LMArena) and the 31B model rivals the recent Qwen 3.5 27B, which is the leading member of that class. The ~30B size range is an important one, as itâs accessible both to researchers and to enterprises looking to deploy the model in real use-cases. Where the 7B model scale is the default for tinkering and research, a 30B model is the default for seeing if an open model can unlock substantial value in your specific workflow â a good mix of intelligence, low price, tractability for downstream training, etc.
This takes us back to the above adoption criteria I mentioned for open models and the bigger question â do I think Gemma 4 will be an overwhelming success? Previous Gemma models have been plagued by tooling issues and poorer performance when being finetuned.
Gemma 4âs success is going to be entirely determined by ease of use, to a point where a 5-10% swing on benchmarks wouldnât matter at all. Itâs strong enough, small enough, with the right license, and from the U.S., so many companies are going to slot it in.
Iâm cautiously optimistic that Gemma 4 is going to work better here. Winds are shifting for open models built in America. We saw GPT-OSS go through a bumpy launch to become an overwhelming success. Thereâs a collective energy around the likes of Reflection, Arcee, Nemotron, Gemma, Olmo, and peers that show substantial demand for building new stacks around open models. Thereâs capital to be spent on AI stacks across the economy by those who want more ownership of everything, including the model.
After launching The ATOM Project 240 days ago, the conversation is shifting into the next stage. Summer of 2025 was a crisis moment where the U.S. AI scene realized it canât wait and figure out open models after building AGI. The two markets will capture different areas and proceed in parallel. Now that more companies in the U.S. are releasing strong models, we need to improve the ecosystem so that these models are easy to use, understand, and build value around. Itâs the hard work to build another inflection point in these adoption plots Iâve been updating consistently, but thatâs the work to be done. Join me in it.
More data coming soon! Hereâs a sneak peek:
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Fast takeoff, the singularity, and recursive self-improvement (RSI) are all top of mind in AI circles these days. There are elements of truth to them in whatâs happening in the AI industry. Two, maybe three, labs are consolidating as an oligopoly with access to the best AI models (and the resources to build the next ones). The AI tools of today are abruptly transforming engineering and research jobs.
AI research is becoming much easier in many ways. The technical problems that need to be solved to scale training large language models even further are formidable. Super-human coding assistants making these approachable is breaking a lot of former claims of what building these things entailed. Together this is setting us up for a year (or more) of rapid progress at the cutting edge of AI.
Weâre also at a time where language models are already extremely good. Theyâre in fact good enough for plenty of extremely valuable knowledge-work tasks. Language models taking another big step is hard to imagine â itâs unclear which tasks theyâre going to master this year outside of code and CLI-based computer-use. There will be some new ones! These capabilities unlock new styles of working thatâll send more ripples through the economy.
These dramatic changes almost make it seem like a foregone conclusion that language models can then just keep accelerating progress on their own. The popular language for this is a recursive self-improvement loop. Early writing on the topic dates back to the 2000s, such as the blog post entirely on the topic from 2008:
Recursion is the sort of thing that happens when you hand the AI the object-level problem of âredesign your own cognitive algorithmsâ.
And slightly earlier, in 2007, Yudkowsky also defined the related idea of a Seed AI in Levels of Organization in General Intelligence:
A seed AI is an AI designed for self-understanding, self-modification, and recursive self-improvement. This has implications both for the functional architectures needed to achieve primitive intelligence, and for the later development of the AI if and when its holonic self-understanding begins to improve. Seed AI is not a workaround that avoids the challenge of general intelligence by bootstrapping from an unintelligent core; seed AI only begins to yield benefits once there is some degree of available intelligence to be utilized. The later consequences of seed AI (such as true recursive self-improvement) only show up after the AI has achieved significant holonic understanding and general intelligence.
Itâs reasonable to think weâre at the start here, with how general and useful todayâs models are.
Generally, RSI can be summarized as when AI can improve itself, the improved version can improve even more efficiently, creating a closed amplification loop that leads to an intelligence explosion, often referred to as the singularity. There are a few assumptions in this. For RSI to occur, it needs to be that:
* The loop is closed. Models can keep improving on themselves and beget more models.
* The loop is self-amplifying. The next models will yield even bigger improvements than the current ones.
* The loop continues to run without losing efficiency. There are not added pieces of friction that make the exponential knee-capped as an early sigmoid.
While I agree that momentous, socially destabilizing changes are coming in the next few years from sustained AI improvements, I expect the trend line of progress to be more linear than exponential when we reflect back. Instead of recursive self-improvement, it will be lossy self-improvement (LSI) â the models become core to the development loop but friction breaks down all the core assumptions of RSI. The more compute and agents you throw at a problem, the more loss and repetition shows up.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
Iâm still a believer that the complexity brake on advanced systems will be a strong counterbalance to the reality that AI models are getting substantially better at every narrow task we need to compose together in making a leading AI model. I quoted this previously in April of 2025 in response to AI 2027.
Microsoft co-founder Paul Allen argued the opposite of accelerating returns, the complexity brake: the more progress science makes towards understanding intelligence, the more difficult it becomes to make additional progress. A study of the number of patents shows that human creativity does not show accelerating returns, but in fact, as suggested by Joseph Tainter in his The Collapse of Complex Societies, a law of diminishing returns. The number of patents per thousand peaked in the period from 1850 to 1900, and has been declining since. The growth of complexity eventually becomes self-limiting, and leads to a widespread âgeneral systems collapseâ.
There are plenty of examples in how models are already trained, the deep intuitions we need to get them right, and the organizations that build them that show where the losses will come from. Building leading language models is incredibly complex, and only becoming more-so. There are a few core frictions in my mind.
1. Automatable research is too narrow
First, it is clear that language models this year will already be useful tools at optimizing localized tasks like lowering the test loss of a model. Andrey Karpathy recently launched his autoresearch that popularized doing just this. This allows AI agents to play directly on GPUs to target tasks like lowering the loss on the test set. This approach works in narrow domains, i.e. one general test loss or one overall reward. The problem is that thereâs a long-standing gap between an on-paper more accurate model and models that users find more productive. The most provocative case is for pretraining, which was discussed more at length around scaling laws. Scaling laws show us that the loss will continue going down, but we donât know if thatâll be economically more valuable.
In post-training, reinforcement learning algorithms are at least more directly tied to specific performance gains as most RL training environments can be used directly as an evaluation. Still, I worry about generalization and tying back to models that are better at the specific task of improving themselves. Itâs a big leap from models get better at some things to that necessarily translating to models that are better at building themselves and designing experiments. Weâve seen many AI capabilities sort of saturate at certain levels of human taste, such as writing quality. AI research is a bit different here, as there is a very high ceiling to climb up to. Where models mostly saturate on writing because thereâs inherent tension in preferences, models will saturate on research because the search space and optimization target is too wide.
The early benchmarks for measuring this sort of ability all fall prey to the same problem â narrow scope. Agents will do well at optimizing single metrics, but the leap required to navigate many metrics at once is a very different skill set. That is actually what the best researchers do â they make many scalable ideas work together.
The most related benchmark we have to measure this is PostTrainBench, which is quite fun, but progress will very rapidly get distorted on this. Over 90% of the challenge in doing post-training well is getting the last 1-3% of performance, especially without cooking the model in out-of-domain tasks. Post-training a general, leading model is extremely complex, and only getting more complex.
I could go on and on about this. Another example is from during my Ph.D. (2017-2022), when there was immense hype around a field called âAutoMLâ which aimed to use techniques like Bayesian Optimization to find new architectures and parameters for models. The hype never translated into changing my job. Language models will do more than this, but not enough to take jobs away from top AI researchers any time soon. The core currency of researchers is still intuition and managing complexity, rather than specific optimization and implementation.
2. Diminishing returns of more AI agents in parallel
The biggest problem for rapid improvement in AI is that even though weâll have 10,000 remote workers in a datacenter, itâll be nearly impossible to channel all of them at one problem. Inherently, especially when the models are still so similar, theyâre sampling from the same distribution of solutions and capabilities while being bottlenecked by human supervision. Adding more agents will have a strict saturation in the amount of marginal performance that can be added â the intuition of the best few researchers (and time to run experiments) will be the final bottleneck.
A common idea to illustrate this is Amdahlâs law, which is taken from computer architecture and shows that a given task can only generate a fixed speedup proportional to how much can be parallelized and how many parallel workers exist. An illustration is below:
In AI this should be relatively easier to convey, as the low-level operating details of computers are fairly mysterious. Consider an AI researcher on the transition from writing code by hand to using AI autocomplete assistance to now using autonomous coding agents. These are all massive gains. Let us continue. Now this researcher uses 3-4 agents working on different sub-tasks or approaches to the problem at hand. This is still a large gain. Now consider this single researcher trying to organize 30-40 agents with tasks to do every day. Some people can get more value out of this scale, but not many.
How many people do you think could come up with 300-400 tasks for AI agents every day? Not many. This problem will hit the AI models soon enough as well.
3. Resource bottlenecks and politics
Fundamentally, all the AI companies are walking a fine line of acquiring substantial capital, converting new compute resources to revenue via sufficient demand, and repeating the process all-the-while spending an extreme amount on research. With the scale of resources here, there will always be political bottlenecks on who gets resources and what gets bet on. In this layer, research leadership sits above the AIs and the researchers. Even as models continue to improve, this source of friction will never get removed. It isnât a substantial friction, but the AI models are fundamentally operating in organizations where humans are the bottleneck on resources.
The early scale of improvements with language models is local optimizations, where the resources used cost
-
Iâm a little late to this model review, but that has given me more time to think about the axes that matter for agents. Traditional benchmarks reduce model performance to a single score of correctness â they always have because that was simple, easy to quickly use to gauge performance, and so on. This is also advice that I give to people trying to build great benchmarks â it needs to reduce to one number that is interpretable. This is likely still going to be true in a year or two, and benchmarks for agents will be better, but for the time being it doesnât really map to what we feel because agentic tasks are all about a mix of correctness, ease of use, speed, and cost. Eventually benchmarks will individually address these.
Where GPT 5.4 feels like another incremental model on some on-paper benchmarks, in practice it feels like a meaningful step in all four of those traits. GPT 5.4 in Codex, always on fast mode and high or extra-high effort, is the first OpenAI agent that feels like it can do a lot of random things you can throw at it.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
I havenât been particularly deep in software engineering over the last few months, so most of my working with agents has been smaller projects (not totally one-off, but small enough where Iâve built the entire thing and manage the design over weeks), data analysis, and research tasks. When you embrace being agent-native, this style of work entails a lot of regular APIs, background packages (like installing and managing LateX binaries, ffmpeg, multimedia conversion tools, etc), git operations, file management, search etc. Prior to GPT 5.4, I always churned off of OpenAIâs agents due to a death by a thousand cuts. It felt like rage quits. Iâd feel like I was getting into GPT 5.2 Codex, but it would fail on a git operation and have me (or Claude) need to reset it. Those hard edges are no longer there.
The other subtle change in GPT 5.4âs approachability â the biggest reason I think OpenAI is much more back in the agent wars â is that it just feels a bit more âright.â I classify this differently to the routine tasks I discussed above, and it has to do with how the product (i.e. the model harness) presents the model outputs, requests, and all that to you the user. It has to do with how easy it is to dive in. This has always been Claudeâs biggest strength in its astronomical growth. Not only has Claude been immensely useful, but it has a charm and entertainment value to it thatâll make new people stick around. GPT 5.4 has a bit of that, but the underlying model strengths of Claude still leave it feeling warmer.
Where Claude is a super smart model, with character, a turn of phrase in a debate, and sometimes forgetting something, OpenAIâs models in Codex feel meticulous, slightly cold, but deeply mechanical. Iâd use Claude for things I need more of an opinion on and GPT 5.4 to churn through an overwhelmingly specific TODO list. The instruction following of GPT 5.4 is so precise that I need to learn to interact with the models differently after spending so much time with Claude. Claude, in some domains, you come to see has an excellent model for your intent. GPT 5.4 just does what you say to do. These are very different philosophies of âwhat will make the best model for an agentâ, Claude will likely appeal to the newcomers, but GPT 5.4 will likely appeal to the master agent coordinator that wants to unleash their AI army on distributed tasks.
Outside of charm, and dare I say taste, a lot of the usability factors are actually better on OpenAIâs half of the world. The Codex app is compelling â I donât always use it, but sometimes I totally love it. I suspect substantial innovation is coming in what these apps look like. Personally, I expect them to eventually look like Slack (when multiple agents need to talk to eachother, under my watch).
OpenAI also natively offers fast mode for their models with a subscription and very large rate limits. Iâve been on the $100/month Claude plan and $200/month ChatGPT plan for quite some time. Iâve never been remotely close to my Codex limits with fast mode and xhigh reasoning effort, where I hit my Claude limits from time to time. Thereâs definitely a modeling reason to this â most of OpenAIâs release blogs showcase each iterative model being substantially more concise in the number of tokens it takes to get peak benchmark performance. This is a measure of reasoning efficiency. This 2D (or more) benchmark picture is exactly where the world is going.
Hereâs a plot from Cursor, which sadly doesnât have all the GPT 5.4 reasoning efforts, but it confirms this point in a third party evaluation. What is missing across model families is the speed and price (a proxy for total compute used) to get there.
The final benefit of GPT 5.4, and OpenAIâs agentic models in general for that matter, is much better context management. In using them regularly now I feel like Iâve never hit the context wall or context anxiety point. The reasoning efficiency I suspect is the case above just lets the model do way more with its initially empty context window. Then, when GPT 5.4 does compact, itâs been less noticeable.
The one problem Iâve been having with both Claude Opus 4.6 and GPT 5.4 is a light forgetfulness. If you give the models multiple TODOs in a single message outside of planning mode, I find them often dropping them. Sometimes it feels like the models glitch and try to solve a previous problem rather than the recent ones. Iâm not sure what in the model or the harness is the exact cause, but sometimes I like to queue up a few messages as I see the model working on something, to refine the task, but currently this tends to be a pretty risky outcome except in the simplest use-cases.
These days Iâve been using both GPT and Claude extensively, mostly based on my mood, and have been getting more done than ever. Having a GPT 5.4 Pro integration directly with Codex, e.g. like \ultrathink, would be a big differentiator for OpenAI. Those models have been incredible.
All in, I see GPT 5.4 as an agentic model that brings a ton more simple usability and âagentnessâ to the very strong software foundation of GPT 5.3 Codex. Itâs a big step, and Iâm unbelievably excited for which of these two companies releases an update next. On paper, listing the strengths of GPT 5.4 across better top end coding performance, better speed, better context management, better rate limits, itâs a testament to how nuanced choosing a model is. I genuinely still enjoy Claude a bit more for ways thatâll never show up on benchmarks. This makes me type claude into my terminal at the start of my day, rather than codex.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
2025 was the year where a lot of companies started to take open models seriously as a path to influence in the extremely valuable AI ecosystem â the adoption of a strategy that was massively accelerated downstream of DeepSeek R1âs breakout success. Most of this is being done as a mission of hope, principle, or generosity.
Very few businesses have a real monetary reason to build open models. Well-cited reasons, such as commoditizing oneâs complements for Metaâs Llama, are hard to follow up on when the cost of participating well is billions of dollars. Still, AI is in such an early phase of technological development, mostly defined by large-scale industrialization and massive scale-out of infrastructure, that having any sort of influence at the cutting edge of AI is seen as a path to immense potential value.
Open models are a very fast way to achieve this, you can obtain substantial usage and mindshare with no enterprise agreements or marketing campaigns â just releasing one good model. Many companies in AI have raised a ton of money built on less.
The hype of open models is simultaneously amplified by the mix of cope, disruptive anticipation, and science fiction that hopes for the world where open models do truly surpass the closed labs. This goal could be an economically catastrophic success for the AI ecosystem, where profits and revenue plummet but the broader balance of power and control of AI models is long-term more stable.
Thereâs a small chance open models win in absolute performance, but it would only be on the back of either a true scientific breakthrough that is somehow kept hidden from the leading labs or the models truly hitting a wall in performance. Both of them are definitely possible, but very unlikely.
It is important to remind yourself that there have been no walls in progress to date and all the top AI researchers we discuss this with constantly explain the low-hanging fruit they see on progress. It may not be recursive self-improvement to the singularity (more on that in a separate post), but large technology companies are on a direct path to building definitionally transformative tools. They are coming.
The balance of power in open vs. closed models
The fair assessment of the open-closed gap is that open models have always been 6-18 months behind the best closed models. It is a remarkable testament to the open labs, operating on far smaller budgets, that this has stayed so stable. Many top analysts like myself are bewildered by the way the gap isnât bigger. Distillation helps a bit in quality, benchmaxing more than closed labs helps perceptions, but the progress of the leading open models is flat out remarkable.
The reality is that the open-closed model gap is more likely to grow than shrink. The top few labs are improving as fast as ever, releasing many great new models, with more on the docket. Many of the most impressive frontier model improvements relative to their open counterparts feel totally unmeasured on public benchmarks.
In a new era of coding agents, the popular method to âcopyâ performance from closed models, distillation, requires more creativity to extract performance â previously, you could use the entire completion from the model to train your student, but now the most important part is the complex RL environments and the prompts to place your agents in them. These are much easier to hide and all the while the Chinese labs leading in open models are always complaining about computational restrictions.
As the leading AI models move into longer-horizon and more specialized tasks, mediated by complex and expensive gate-keepers in the U.S. economy (e.g. legal or healthcare systems), I expect large gaps in performance to appear. Coding can largely be mostly âsolvedâ with careful data processes, scraping GitHub, and clever environments. The economies of scale and foci of training are moving into domains that are not on the public web, so they are far harder to replicate than early language models.
Developing frontier AI models today is more defined by stacking medium to small wins, unlocked by infrastructure, across time. This rewards organizations that can expand scope while maintaining quality, which is extremely expensive.
All of these dynamics together create a business landscape for open models that is hard to parse. Through 2026, closed models are going to take leaps and bounds in performance in directions that it is unlikely for open models to follow. This sets us up for a world where we need to consider, fund, use, and discuss open models differently. This piece lays out how open models are changing. It is a future thatâll be clearly defined by three classes of models.
* True (closed) frontier models. These will drive the strongest knowledge work and coding agents. They will be truly remarkable tools that force us to reconsider our relationship to work.
* Open frontier models. These will be the best open-weight, large models that are attempting to compete on the same directions as above. There will be plenty of use-cases that they donât work for relative to the best models, but countless use-cases where they work remarkably well. For many use-cases, even ones as valuable as some subsets of coding, these will work great. The AI ecosystem will still take years to understand what it means to have intelligence of this magnitude served in private, at the marginal cost of electricity for individuals, as assistants, coaches, companions, and more. OpenClaw provided a glimpse behind the mirror that will expand and grow. The class of models around GPT-OSS 120B, Nvidia Nemotron 3 Super, or MiniMax M2.5 are the balance of performance to price that can work as local models.
* Open, small models as distributed intelligence. The most successful open models will be complementary tools to closed agents. This is a path for open models to complement and accelerate the frontier of progress.AI is slotting in to automate many repetitive, niche tasks across the technology economy. Thereâs a huge pressure to shift these tasks off of the best closed models â which frankly are still better at most of the things, across my conversations with businesses trying to build with open models â to small, open models that can be 10X faster and 100X cheaper. There arenât really people building data and fine-tuning engines for economically viable tasks on the smallest models possible. These models need to be almost brain-numbingly boring and specific. In a world dominated by coding agents, I want to build open models that Claude Code is desperate to use as a tool, letting its sub agents unlock entirely new areas of work. This is possible, but remarkably under-explored. Small models from the likes of Qwen and co. are still marketed on general-task benchmarks. The hype of âopen models catching the frontierâ distracts the world from this very large area of demand.This is the sort of model that moves open models from just a few, crucial static weights to more of an ecosystem. It requires creativity and a new approach. The goal of this piece is to illustrate why and how to build these, with added context on where open models stand today.
All three of these model classes hint at different ways to use agents. It is absolutely definitional to how AI is going to be built going forward that theyâre not just model weights, but rather systems that think, search, and act. The weights only define one portion of those abilities.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
Open weights as part of an AI system
To start, consider what are the most impactful and impressive things that language models can do without a suite of tools at their side. When was the last time that you were blown away by something that was just autoregressive token outputs? Unless youâre doing a substantial amount of work on mathematical proofs or competition code, it seems like that situation has changed little since GPT-4âs release in 2023. The AI systems we use today are about far, far more than weights.
In this world, closed models have a clear advantage. Closed models get to vertically integrate everything from the chips they run on, the inference software, the weights, the tools, and the user interface. Open models on the other hand need to work on every inference setup, with many tools, and in many use-cases. This vertical integration is best expressed today in the joy of using Claude Code with Opus 4.6 or OpenAIâs Codex with GPT 5.4. Open models havenât passed this point. Some are starting to focus on specific interfaces, e.g. OpenCode, but thereâs an inherent tension in making an open model work only in your blessed product roadmap.
At the same time, this change could point to more about the latest AI systems being open! If you can do less with the weights alone, maybe more labs will release them.
The way to think about AI systems today is as a mix of weights, tools, and harnesses. The weights portion is familiar. The tools are the deeply integrated environments the models act in at deployment time â best typified by search and code sandboxes â and the harness is how these two fit together with a product that the user sees.
In this world, there are two things to consider: 1) Is there an equivalent, open system to the closed products that people are using today â I mean truly equivalent, where every level of the stack can be modified and controlled (more on this later), and 2) How does this systemâs view impact different future decisions in the open ecosystem?
Still looking for open model business strategies
To understand how the business and practicality of open models will evolve, let me take a tour back in time to foundational writing on the role of open-source in modern technology companies. The first is a Google blog post, The Meaning of Open, which originally was an internal memo by Jonathan Rosenberg, which sparked an intense internal debate that later resulted in it becoming public. To start, hereâs a basic assessment of how open systems can work:
Open systems have the potential to spawn industries. They harness the intellect of the general population and spur businesses to compete, innovate, and win based on the merits of their products and not just the brilliance of their business tactics.
Iâve long believed that the company who will benefit most from the ecosystem of open models is the one who understands it best. This entails being deeply involved with open research and experimentation in how to use the models. So far, most of the open model company business models are not this. Rosenberg expands on this in his 2009 post, comparing the dynamics of open systems to closed products:
[Open systems] are competitive and far more dynamic. In an open system, a competitive advantage doesnât derive from locking in customers, but rather from understanding the fast-moving system better than anyone else and using that knowledge to generate better, more innovative products. The successful company in an open system is both a fast innovator and a thought leader; the brand value of thought leadership attracts customers and then fast innovation keeps them. This isnât easy â far from it â but fast companies have nothing to fear, and when they are successful they can generate great shareholder value.
Weâve known for some time that open weight models are not actually enough to constitute a product â models are a product in the sense that they have tools and harnesses, so we donât actually have fully open systems, we have systems that are partially open partially closed, making moats messy. VLLM and a model like GLM 5 are pieces of a system, but it still takes more to deploy them â expensive private GPUs and some tools with local business data.
It may turn out to be that AI is too complex and expensive to have any analogous open system to previous generations of technology. If there was a fully open system, it would win by default, as many historical generations of technology have shown us. This fully open analog does not yet exist, so we have constant debates on the role of open-source AI.
Bill Gurley recounts how Googleâs free products have exemplified the open or free strategies across technology. Gurley wrote on the open-source operating system, Android, and the free browser, Chrome, in 2011:
So here is the kicker. Android, as well as Chrome and Chrome OS for that matter, are not âproductsâ in the classic business sense. They have no plan to become their own âeconomic castles.â Rather they are very expensive and very aggressive âmoats,â funded by the height and magnitude of Googleâs castle. Googleâs aim is defensive not offensive. They are not trying to make a profit on Android or Chrome. They want to take any layer that lives between themselves and the consumer and make it free (or even less than free).
Because these layers are basically software products with no variable costs, this is a very viable defensive strategy. In essence, they are not just building a moat; Google is also scorching the earth for 250 miles around the outside of the castle to ensure no one can approach it.
In the same post, Gurley reflects on the limits of Googleâs openness:
In this open manifesto, Jonathan opines over and over again that open systems unquestionably result in the very best solutions for end customers. That is with one exception. âIn many cases, most notably our search and ads products, opening up the code would not contribute to these goals and would actually hurt users.â As Rodney Dangerfield said in Caddyshack, âIt looks good on you, though.â
Essentially, Google open-sourced so much, in fact paid people to use its products (e.g. paying phone makers to use android) to keep the funnel leading to the search profit center. This is the virtuous loop that the search business still funds to this day.
AI is still nothing like this, but signs of change are emerging. The default belief on the value of models to these companies is that the model is the product. This is obvious with products like hosted APIs, where releasing the model weights would be business suicide, but this is softening as interfaces like Claude Code, Codex, Cursor, etc. get vastly popular. It could be a path to more openness, at least in parts of the stack. We can see this with the coding plans offered by Moonshot and Z.ai â where the demand is very high for the businesses, even though the model is open. Most people will just use the cheap interface with inference, instead of figuring out how to use the model themselves (as long as the business is mostly consumer or per-head services).
All of this doesnât leave me optimistic on the direction of companies becoming more open in the coming years. Iâd expect the opposite still. Nvidia has the one great reason to be open â to sell more GPUs to people building on open models and understand what they need to build next, but thereâs no one else obvious on this list. Until there are more specific economic reasons to build open models, the companies building these at the frontier will have fewer resources to spend on the models and face a consolidation to the best few.
In the face of consolidation at the open frontier, the investment in the models should shift to areas where the models can have more differentiated upside relative to the best closed frontier models.
Open models that are specific, cheap, fast, and ubiquitous
Thereâs too much obsession with the best companies building open models to try and compete at the frontier. Thereâs a vastly underserved market of enterprises that want cheap, reliable models for repetitive use-cases in their systems. Picture this, one small model with a series of LoRA adapters that specialize the model to internal skills. This can be deployed very cheaply as tools and a complement to the frontier closed models that are orchestrating agents.
Every task that a frontier agentic model does tens to hundreds of times can potentially be outsourced to a small model. There are ancillary benefits to this, e.g. privacy of a local model reading your files and summarizing to Claude, but almost no one is pushing hard in this direction. The leading model family of capable, customizable small models to date is Qwen, but thatâs now shrouded in uncertainty with the departures of key personnel. Gemma, Phi, Olmo, etc. are all major steps down in quality, and therefore potential for modification.
There are a few obvious examples why this can be scaled up. There was a recent thread and discussion on how the new Qwen 3.5 4B model arguably bests the original ChatGPT model. On the research side, there are already recipes for finetuning open models on specific code-bases to match performance of much bigger models. Moondream.ai is a startup made by a friend of mine Vik, who builds some of the best, small multimodal models on a tiny budget â they compete with Qwen and Llama on real world tasks. This is the tip of an iceberg.
Intelligence compression hasnât been explored with nearly as much depth (or resources) because it is less exciting than keeping track of the progress of the best few models. Investigating these areas is the standard technological diffusion process that is slow and why weâre still early in understanding how people will build with AI. My contention is that too many people building open models are slightly deluded in their perception of their competitiveness. The best few models will win on general capabilities and there are still plenty of underserved niches elsewhere.
Taking this to the next level involves releasing open models that are scoped to be truly excellent at 1-3 tasks, as I hinted at the beginning of this piece. Too many people try to compete with Qwen and show that their small model does great on frontier AI benchmarks. The right benchmark here is savings in compute and time.
Itâll take years for this transition to slowly become reality. Part of why I am so excited about it is that it is driving innovation on open models being more about diversity, specialization, and curiosity, rather than the standard âone model to rule them allâ that the frontier models presume.
Models vs. ecosystems.Consolidation vs. creativity.
So long as the open source ecosystem for AI is defined by a bunch of model providers trying to chase after the closed labs, it will largely lose. It will face pain on funding and substantive adoption. The same consolidation that will come for closed AI companies will come for open model builders â likely even sooner.
Open systems at their best allow many people to participate and many approaches to flourish.
The world of open models needs to be more of an ecosystem. Iâve discussed in the past how China is closer to this type of environment by having a variety of companies, but the variety in approaches is still too low.
Ecosystems are self-reinforcing, whereas individual models are static artifacts in time. Ecosystems showcase clear, constant opportunities for whatâs next that have growing value propositions.
The path forward for open models is to solve different problems than the frontier labs, to find places where open models are effectively free alternatives, to show ways of using specialized models that the closed labs cannot offer. The world of open models needs to embrace creativity, before building powerful AI systems grows too expensive and prices out many of the prized open labs of today.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Watching history unfold between Anthropic and the Department of War (DoW) it has been obvious to me that this could be a major turning point in perspectives on open models, but one thatâll take years to be obvious. As AI becomes more powerful, existing power structures will grapple with their roles relative to existing companies. Some in open models frame this as ânot your weights, not your brain,â but it points to a much bigger problem when governments realize this.
If AI is the most powerful technology, why would any global entity let a single U.S. company (or government) control their relationship to it?
I got Dean W. Ball of the great Hyperdimensional newsletter onto the SAIL Media weekly Substack live to discuss this. In the end, we agree that the recent actions by the DoW â especially the designation of Anthropic as a supply chain risk (which Dean and I both vehemently disagree with) â points to open models being the 5-10 year stable equilibrium for power centers.
The point of this discussion is:
* Why do open models avoid some of the power struggles weâve seen play out last week?
* How do we bridge short term headwinds for open models towards long-term strength?
* The general balance of capabilities between open and closed models.
Personally, I feel the need to build open models more than ever and am happy to see more constituencies wake up to it. What I donât know is how to fund and organize that. Commoditizing oneâs compliments is a valid strategy, but it starts to break down when AI models cost closer to a trillion dollars than a hundred million. With open models being very hard to monetize, thereâs a bumpy road ahead for figuring out who builds these models in face of real business growth elsewhere in the AI stack.
Enjoy and please share any feedback you have on this tricky topic!
Listen on Apple Podcasts, Spotify, and where ever you get your podcasts. For other Interconnects interviews, go here.
Chapters
* 00:00 Intro: is the Anthropic supply chain risk good or bad for open models?
* 04:03 Funding open models and the widening frontier gap
* 12:33 Sovereign AI and global demand for alternatives
* 20:55 Open model ecosystem: Qwen, usability, and short-term outlook
* 28:20 Government power, nationalization risk, and financializing compute
Transcript
00:00:00 Nathan Lambert: Okay. We are live and people will start joining. Iâm very happy to catch up with Dean. I think as we were setting this up, the news has been breaking that the official supply chain risk designation was filed. This is not a live reaction to that. If we get any really, really interesting news, weâll talk about it. I think one of the undercurrents that Iâve felt that this week where everything happened is gonna touch on is open models, but thereâs not an obvious angle. I think I will frame this to Dean to start, which is how does-- Like, thereâs two sides of open models. One is that thereâs the kind of cliche like, not my weights, not your weights, not your mind, where like somebody could take it away if not an open model, which people are boosting like, âOh, like Anthropicâs gonna take away their intelligence.â But the other side is people worried about open models existing that the Department of War can just take and use for any purpose that it wants. And I feel like both of these are a little cliche. And the core question is like, is this type of event where more control is coming towards AI and more multi-party interest, like is that gonna be good or bad for the open weight model ecosystem?
00:01:12 Dean Ball: My guess is that in the long run, this is probably profoundly good for open weight AI. And like the whole reason I got in, like, so I became interested in frontier AI governance. I did something totally different with my time before. I wrote about different kinds of policy and studied different kinds of policy. And the reason I got into this was because it immediately occurred to me that the government was gonna... I was like, okay, letâs assume weâre building super intelligence soon or whatever, like very advanced AI that seems like really important and powerful. Thatâs gonna be something that I depend on, like for my day-to-day life. Iâm gonna need it for all kinds of things. Itâs gonna profoundly implicate my freedom of expression as an American and my exercise of my liberty and all that. And yet itâs also gonna profoundly implicate national security. And so the governmentâs gonna have its hands all over it, and they also might not like me using it because I might use it, and others might use it to challenge the status quo in various ways, to challenge the existing power structures which the government is a part of. So we have a political problem on our hands here, in my view.
00:02:36 Dean Ball: It immediately occurred to me that weâre gonna have this huge problem of like, this is gonna be a conflict because this is something thatâs gonna enormously implicate American speech and liberty, and also itâs gonna have legitimate national security issues, and also the governmentâs gonna want it because of bad power-seeking reasons. And so thatâs always a part of the picture. And my view was this is just a fight thatâs gonna play out over the coming decades, and I wanna be a part of this fight. But number two, in that fight, you have to have an insurance policy, and open weight is the insurance policy. Open weight is the way we can always say yes, but we can build the open ecosystem. We can do that. And so I think in the fullness of time, this is gonna be beneficial, but the problem is thereâs a lot of coordination and economic problems that have to be solved here. Itâs not just a matter of hoping that Google and Meta or whomever else, or the Chinese companies, by virtue, out of the goodness of their hearts continue to open-source things. Thatâs not scalable. There has to be a reason to do it. So what are the institutional dynamics open weight gonna look like in the long term? I donât really know, but it feels deeply under theorized.
00:04:03 Nathan Lambert: I think itâs hard to fund is the thing. I mean, we saw Qwen had their turmoil this week, which is timely, and Iâm not that surprised because the stakes for these companies is so high, and they all are trying to make sure their companies win in it. And people will say like, âOh, Meta should commoditize their complements and release open models.â But no oneâs ever commoditized their complements with something that costs a trillion dollars to make. Like, thatâs a line item. Like, is Apple gonna commoditize... Apple commoditizing their complement would be them doing the... They could spend just as much as all the other tech companies are on CapEx and spend hundreds of billions of dollars, but theyâre choosing not to. And I just like, I agree that long term it should be better, but if we never bridge that gap, does it actually materialize? Like, the crank is being turned of these models getting better and better. GPT 5.4 released today, excited to try it.
00:05:02 Nathan Lambert: But like, where does it go? Like, what Iâm working on is totally falling behind the frontier. Weâre the foundation of research, but itâs like I see it already slipping.
00:05:13 Dean Ball: So I kinda think, yeah, I mean, look, I think itâs gonna get bad in the short term, itâs gonna be bleak, right? Thereâs just no doubt about that in my view. Because weâre in this period, like I think the pace of frontier progress is gonna continue. My own view is that, like, just âcause I peer in and use the open weight Chinese models on a fairly regular basis, and I kinda just feel as though the gap has widened between the US frontier and the open frontier. Unfortunately, itâs so sad that US frontier and open frontier are increasingly distinct things. But I do feel as though that probably is true. And thatâs probably gonna continue because in the next, like, in the early stages of a new technology, you would expect for the vertically integrated players to be the ones who do the best. And over time, the modular players can win, and part of that is âcause eventually you do get to good enough, right? Like, eventually, I think most people think the iPhone is good enough now. There was a time when every year the iPhone upgrade was like, âOh my God, this is so much better.â Intelligence is maybe different, but maybe not for a lot of things.
00:06:37 Nathan Lambert: Well, like, thereâs no iPhone that you can buy from anyone. Nothing you can buy from anyone but Apple is nearly as good. Thatâs the concern. Itâs like, is it gonna be Anthropic that like, yeah, it stopped getting better, but you canât rebuild it. Like, you canât make the open source version.
00:06:51 Nathan Lambert: I also think I had a later question, which is like, the weights are so much less of a concern for me. So like, somebody dropping a two-trillion-parameter model thatâs open weights and way better than anything else that somebody has built and released in the open, it almost doesnât matter if you donât understand the harness and the tools and the setup you need to make it into a Claude-like system. Like, you need what, eighty nodes of H100s that cost a hundred thousand dollars a day to run and expertise to make it a system. Itâs like the shifting away from weights is also happening. I donât think itâs happening in this open versus closed ecosystem at the surface level of the discussion. So thatâs why Iâm just like, I donât know if itâs gonna exist. The thing that I could see happening is that open weights models are niche, and they help these Claude-like models, but thereâs not an alternative in that universe. So itâs like, is the government capable of actually making this alternative exist? I donât know. Like, I donât know if you can Manhattan Project this, and I wouldnât advocate for it.
00:07:53 Dean Ball: I actually think about it from the opposite perspective, because I think that what happens if the government follows through on what theyâve threatened with Anthropic, which is to make it so that basically any military contractor cannot have any commercial relations with Anthropic, which means NVIDIA canât sell GPUs to them for anything. Amazon canât sell cloud services to them. Amazon and NVIDIA also canât be invested in them, by the way, if you take any commercial relations at its face value. Now, thatâs not a power the government actually has, but nonetheless, if this harassment campaign continues, I think what it probably does... You know, I spend a lot of time in international policy, dealing, talking to foreign governments and civil society in foreign countries, and they already have major trust issues with respect to the US closed source models because they think the US government is gonna come in and disable the models. Like, the American president will get mad at Brazil, say, and in addition to putting tariffs or sanctions, the US president will say, âYeah, weâre also gonna turn off all your public services that are dependent upon American closed source models.â Right? So people view that as this profound threat, and people are legitimately scared of that in other countries.
00:10:00 Dean Ball: I think this turns that fear up another meaningful degree, and probably not incorrectly, by the way, probably rightfully so. And so I kinda look at this and I think, well, now a lot of American companies might also have that concern, and so you certainly have a demand side of people who are gonna be like, âI get this. It is a risk to use anything where I have a commercial relationship. âCause once I have a commercial relationship, the government can regulate that. Can I find some way of getting out of it?â I think thereâs gonna be demand for that. Whether or not that demand produces supply, I think will depend on... It might just not be possible, thatâs true. But I think youâve never had a more favorable demand picture, and I suspect that on the margin, this probably will favor open in the longer run.
00:10:44 Nathan Lambert: Yeah. So thereâs a few ways that I think about this. I have this thing, like ATOM Project and all this other stuff I do, and itâs like, how do I meaningfully advocate for this? I think thereâs something, like I work at AI2, and AI2 has budgets of order of a hundred million dollars and can train decent models. But if I wanted to redo an AI2, like my method for getting that type of money, itâs mostly gonna be like befriending a billionaire. And it seems like philanthropy dice roll in the near term is a way to get it. But then, like, maybe it really is some long slog of a multi-industrial consortium that takes a couple years off the ground and slowly, like, Googleâs, or all these Netflix and all these five hundred billion dollar smaller companies are gonna give millions of dollars to have somebody else do it because they canât get the billion dollars themselves, but they know they need to have it existed.
00:11:31 Dean Ball: And sovereign wealth funds. Right. Sovereign wealth funds everywhere can do that, right? Thereâs trillions of dollars in sovereign wealth. Thereâs pension funds, public employee pension funds. A lot of people can chip into this and itâs possible. This is like, Yann LeCun thinks this is the inevitable outcome. He thinks that the future is gonna be that some sort of global consortium gets together and builds this, because no one country is gonna be able to own it, because itâs gonna be too important. Iâve always kinda doubted that, and Iâve always thought that that outcome is probably a bad outcome for the world, honestly.
00:12:06 Nathan Lambert: Thatâs a bad outcome for how good the AI is.
00:12:09 Dean Ball: Thatâs correct. Itâs a socialist outcome, you know? Itâs not communism, but it is democratic socialism, and Iâm not a democratic socialist, so Iâm not a super big fan of that. But at the same time, I have to be honest that I kinda think that this probably does increase the odds of that precise outcome coming to bear.
00:12:33 Nathan Lambert: I think something that comes sooner is that a lot of these super wealthy countries are gonna realize they can have real... Like, they can do some sort of sovereign AI and make some sort of noise, particularly starting with open models. I think thereâs the Institute for Foundation Models, which is based on the UAE university system. Like, thatâs--
00:12:53 Dean Ball: Thatâs very UAE-coded, yeah.
00:12:55 Nathan Lambert: Theyâve been playing that for years, and they can keep doing this. Their models are gonna be pretty good, and I think thereâs gonna be more people that do this. Thereâs the SWISS initiative in EU, which is on one hand doing a good job, on the other hand plagued by the most obvious European limitations of talent cycling and consortium life. I think these things are gonna become more of a thing in the next year, but I donât know exactly how they impact the... They donât impact the frontier of AI, but maybe theyâre just like how the geopolitics and power of AI evolves. And I for some reason feel like open models need to be the thing that theyâre gonna do because if they have a closed model thatâs not as good, it doesnât really give them any sort of power. But I donât have a good enough world view for what that actually does, and if thereâs more EU models, if India actually has their act together and trains a solid model. I donât know what that does, but I feel like itâs probably gonna happen.
00:13:54 Dean Ball: Yeah. I mean, itâs really super interesting âcause I think the other thing-- that will be inherently... I mean, it will be a Linux compared to a macOS, you know? It will not be as good of an experience for people. But then it becomes strange. Like, I donât think macOS is as appealing of a thing if itâs viewed to be owned by the US government, right? And in fact, part of the reason I think that Apple is able to make its case quite credibly to consumers and businesses is they have resisted US government pressure to turn things over before. People might remember about a decade ago, there was this shooter in San Bernardino, California, and the FBI tried to force Apple to release iPhone data, and Apple said, âNo, weâre not gonna expose this information.â Now, I think the FBI eventually just hacked it anyway, but thatâs a separate issue. Itâs a matter of principle here.
00:15:01 Dean Ball: So yeah, I think itâs an interesting question: do we expect for the gap between the open frontier and the American closed frontier to widen in the near future, especially just because of how much compute theyâre gonna have?
00:15:30 Nathan Lambert: A hundred percent. And data and talent. Like, a hundred percent. Itâs happening.
00:15:34 Dean Ball: Data, talent. And itâs compounding, right? I mean, this has always been my view. And how much, Iâm not sure, but I think it could be quite significant because these things are compounding benefits. And so if you expect them to just continue compounding, then all of a sudden it gets pretty bleak pretty quickly, would be my fear.
00:16:00 Nathan Lambert: One of the... I mean, whatâs your take on this? Why has it not compounded so much faster? Like, I feel like these three companies are spending, I donât know, 10X what the Chinese labs are spending, and you only get like a little bit better model. Like, I believed so full-heartedly that Claude and ChatGPT and all these models are much better, and I expect them to become better by increasing margin, but itâs still confusing why theyâre not already more ahead.
00:16:29 Dean Ball: I go back and forth on this. Sometimes I think they are that ahead, and itâs just difficult to show up in benchmarks for the obvious reasons that benchmarks get chased. And like, I do feel that with the coding agents and with certain use cases, I do just feel like, wow, the American frontier is just way ahead, profoundly ahead of the Chinese frontier there. But thereâs a lot of other things where you do kinda saturate how good you can be. I suspect that a very large fraction of AI usage is essentially glorified Google search. Even though I donât think AI is glorified Google search, I suspect that a lot of what people use it for is that, at the consumer level. And it isnât obvious to me how much better you can get at things like that. But my guess would be that over the next five years, I would guess the American labs really take off, in part because of compute, data, internal deployments for recursive self-improvement style stuff. And also, itâs amazing how we talk about that as just a normal thing now.
00:18:05 Nathan Lambert: I think there will be a ceiling on it. Like, theyâre gonna get a ton of improvement-- The gains are insane. Itâs like, personally, at my job, Iâve been a lot of a research manager and just chasing s**t down to get a model out the door. But now I can take on hard engineering tasks because Iâm like, âOkay, might as well do this at the same time.â Like, going from zero to a hundred software engineers at anyoneâs fingertips is worth a lot in terms of exploration. But the next, like, from a hundred to ten thousand is like, people can mess that up type thing. But thatâs a huge gain.
00:18:37 Dean Ball: I kind of agree. I think thereâll be a sigmoid there too. But then the other thing that will happen is, like, what I sort of wonder is will the AI companies, will the current model vendors, will they eventually become more like true infrastructure companies where what they actually do is they have models that design their own chips and models that design their own data centers and models that design their own successors. And so itâs this hugely vertically integrated thing, and what youâre really getting access to is not just the model itself, but youâre getting access to this highly optimized hardware, physical world infrastructure. And again, thatâs kind of already the case, but does that become even more the case? And then thatâs truly insurmountable for any open player. Thatâs definitionally insurmountable for an open player, and that becomes scary too. But again, this is why Iâve always felt so good about the position of the US closed source labs. This is why Iâve always been pretty bullish on them and have my concerns about open.
00:20:07 Dean Ball: But to the extent the US government makes it impossible to trust closed source models, you do provide an advantage to open there. Youâre giving a shot in the arm. If you like open source, you should hope that the supply chain risk designation against Anthropic is quite broad.
00:20:09 Nathan Lambert: Itâs a rough thing to hope for.
00:20:09 Dean Ball: I mean, you shouldnât actually hope for it, but I just mean, like, if thatâs the only thing you care about in the world is open source, then--
00:20:17 Nathan Lambert: I would say that anyone that only cares about open source probably is not thinking through any of these principles. It just gets really bad if you only have-- Like, AI is not gonna be meaningful lift to the economy and nor sustainable if everything is open. Like, if models are truly commoditized, things look kind of rough out there.
00:20:36 Dean Ball: I think a world where models get commoditized is a really bleak world too, actually. And yeah, this is why Iâm very worried about what the US government is doing. But I think that it helps on the margin, though. It probably helps on the margin in terms of waking people up. That still is my view.
00:20:55 Nathan Lambert: I am a little surprised by the Qwen stuff, but I think thereâs-- Itâs like, at some point, I knew there was gonna be a year where a lot of the open model efforts just died because theyâre just too expensive and too similar. But at the same time, having a lot of efforts that are somewhat similar but exploring a lot of the minor permutations in modeling space to figure out what works for people who use open models is actually quite good. Iâm very bearish on the reflection style approach, which is build a lab, build an incredible model, drop it, make a bank selling it on-prem. Because on-prem is not that distinct from a business model as having a closed model. You could sell a closed model on-prem with the right IP controls. But then the person who actually wins open is by trying a whole bunch of tiny different things, understanding what is actually a meaningful differentiator in private data, in certain deployments and whatever, and then really iterating on that with a community. And thatâs why I was like, Qwen is the closest to doing this by being so close to the community, and itâs so distinct from what a lot of the other labs are betting on.
00:22:05 Nathan Lambert: But I see the pressure going away and kind of reducing diversity onto standards, because standards also make inference more efficient. Using open models is really rough. I think some of the best open models have really had rough launches. I think GPT-OSS had a horrible launch in terms of usability and is now one of the most popular models of all time. Qwen 3.5, itâs like researchers I work with are like, âOh, letâs see if we can do some basic RL baselines on it,â and all the software stack is kinda broken. It takes a few weeks to get it going. And this is âcause all the models change differently, and closed labs just have such an advantage there âcause they should conceivably ship things on day one that work. I mean, donât talk about Claudeâs runtime, but thatâs fine.
00:22:42 Dean Ball: And donât talk about the GPT-5 auto router either. But yeah, no, totally. I think thatâs right.
00:22:53 Dean Ball: I think fullness of time, Iâm bullish on open source in the long run, fairly bearish in the next five years. The next five years are gonna matter quite a bit. And there is a lot of cope in both open source world and also... I donât really hear it so much in open source world. I think open source world is actually more honest about this. But where the cope is so bad is in global civil society discourse. Like, I was in India for the AI Impact Summit recently, and they are just smoking the copium, being like, âWe are gonna do everything on subfrontier open source models, and weâre just gonna diffuse those, and thatâs all weâre gonna need in our economy.â And I just think thatâs, if youâre India, thatâs really not the bet you wanna make. I understand these are resource-constrained countries. They have a lot of acute constraints that they face, but nonetheless, I think thatâs probably not a good bet.
00:24:05 Nathan Lambert: Well, itâs even if those long tail models will work like manufacturing has worked, where itâs like Apple has put hundreds of billions of dollars into the manufacturing ecosystem in China to get absolute fine margins and scale. Like, if you really-- these things are gonna be used so much that that fine margin is actually gonna matter a lot, and it is not cheap to get that fine margin. You canât just YOLO a DeepSeek V3 and spend five million dollars in compute and be done. Itâs still gonna be expensive for a long time.
00:24:34 Dean Ball: Yeah, it requires-- I think the Chinese approach, in the long run, if Chinaâs gonna continue its strategy and they want to be competitive with the American frontier, theyâre gonna have to fully socialize that, I think. I donât think DeepSeek alone is gonna be able to do this, and I donât think even Alibaba alone is gonna be able to do this. I think theyâre going to need some sort of collective effort. Especially because of the export controls, the American export controls. Theyâre gonna have to centralize compute. Theyâre gonna have to centralize all these things, and talent and data and all that.
00:25:17 Nathan Lambert: I donât see it happening. Like, maybe someone gets officially AGI pilled, and I donât know that much about China. But the things I know about China, it seems like that would be a big lift, and it would take a lot of time to actually do it. Like, all the companies would have to give up their biggest... All the cloud companies are like tech companies making a lot of money. They would be like, âWe have to give up what?â
00:25:42 Dean Ball: No, it would be a tough sell. Obviously, if the Chinese government decides they want to do it, they absolutely will. But in total, it will be a tough sell. My experience having had diplomatic engagements of many sorts with Chinese government-- and a lot of Chinese tech policy is actually not directly set by the government. Itâs actually more kind of civil society, academia and civil society adjacent to government. Had a lot of conversations with folks like that, and theyâre definitely... Itâs largely not a very AGI-pilled crew. I think AGI-pilled-ness probably has a rough correlation with GDP per capita, and I think China is about where you would expect based on their GDP per capita, maybe a little bit ahead, but not very so. But if they ever do get AGI pilled, thatâs the kind of thing that they could consider, but then thatâs still a pretty extraordinary outcome because the Chinese government would have to be willing to make these things and then give it away. And I kinda just donât think they will.
00:27:11 Nathan Lambert: Yeah. I mean, all the politics of control with how everybody thinks AI is so powerful are pointing to very value-destructive actions economically in order to achieve the end state that people determine to be right. Itâs like supporting open source to the extent that you can to avoid situations like Anthropic being labeled a supply chain risk and having interactions like that totally decimating runway of AI productivity. Like, if the companies are really gonna commit to open source for other things, then theyâre gonna lose money. And I see this in-- Chinaâs economy would be taking a gigantic hit doing this. And thatâs kind of a common theme of what weâre talking about is that the interface of AI in an economic fashion is gonna make the next few years really weird.
00:28:06 Dean Ball: I hope so.
00:28:09 Nathan Lambert: I think things are gonna be weird, but I havenât spent a ton of time thinking about how that interacts with political institutions. I thought about socially weird a lot, but I havenât thought about power weird a lot.
00:28:20 Dean Ball: Oh, power weird is what I worry about all the time. What I worry about the most is I think itâs plausible that what weâre seeing... Iâve always had this concern. I have this dual problem of-- maybe Iâm talking out of both sides of my mouth. Maybe thatâs just the critique, and itâs a fair critique. But I routinely complain about how people in government arenât really... They pretend to take AI seriously, but they donât take it that seriously. And they donât really own the implications of advanced, of near term advanced AI and all that. I think we basically have transformative AI right now, but they donât own that, because itâs annoying, itâs difficult, itâs conceptually challenging.
00:29:08 Dean Ball: But the flip side of that is that if people do start to take it very seriously, thereâs the risk that they sort of lash out, that they get scared, and they lash out and do things that are rash, in a rush. And that actually creates very, very bad, much worse outcomes than you otherwise might have gotten. I think thatâs a very fair risk, and I think itâs possible that you might see things like that happen within the U.S. I donât think this particular incident with Anthropic is quite an example of that. But itâs possible that you do see that in the coming years, and that is in and of itself a pretty scary outcome because if the U.S. government decides that they want to nationalize the frontier labs, I think it could be one of the most tyrannical things we ever see happen in this country.
00:30:16 Nathan Lambert: Yeah. Itâs like, I donât know how to reply to this. I think things are... Itâs serious times and I see so many... It feels like such a Sisyphean task to make more open models exist, but all the broader trends seem to point to that being a more stable equilibrium in a lot of ways. Like, good enough open models and keeping up with what we all feel happening in the closed model land.
00:30:50 Nathan Lambert: So I donât know. I stay motivated, but I feel increasingly lost in terms of achieving it.
00:30:56 Dean Ball: I donât think you should be. I think, look, I suspect the US government will not actually do it, and the best thing about America is that our general sort of-- I donât wanna say incompetence, but the general sort of chaos of American institutions and decentralized confusingness of it all, it can often be quite frustrating, and it can sometimes be a detriment, but it can also be really great because we tend to not execute and follow through on our very worst ideas. And so I donât think weâre going to do that. It doesnât feel very American to do it. I worry about it because I worry about these rash reactions, and thatâs why I fight as heavily as I do on things like this, despite not insignificant cost to me to do it, politically speaking. But thatâs totally worth it because I care about this. I think everything, I think that will probably be fine. But yeah, I do agree. Itâs a major risk. Itâs a major risk, and itâs a weird world to think about, Iâll tell you that much.
00:32:16 Nathan Lambert: Yeah. I donât have a lot more to add. Iâm sure weâll continue this discussion. I think it warrants the space of it âcause thatâs the... Itâs one of the longer term things, but itâs not in the news cycle whatsoever, at least the open model angle. Thereâs just so many layers. People have to talk. Like, send feedback, people listening. Iâll even send this out as a podcast as well and just like, what do people think? How do we get to the places we want to get to?
00:32:46 Dean Ball: Well, one thing Iâm particularly interested in is-- one of the items in the Trump administration action plan, which I worked on for those who donât have that context, is this idea of financializing compute, creating a financial market, like basically a commodities market for compute so that you can buy, you know, like really robust. In the same way that you can buy electricity spot, electricity futures and electricity on the spot market and things like this, the wholesale. Could you do something like that for compute? That could really profoundly change the dynamics and the economics of AI production. Itâs not gonna turn them over. It doesnât flip them on their head, but it changes it quite meaningfully. And Iâm very excited by that prospect.
00:33:48 Dean Ball: And thatâs the kind of thing that I would be increasingly doing if this sort of interference of government into the frontier continues. What I suspect Iâll do is start developing some of those ideas which I developed earlier. Iâm only one person. If those things start to seem relevant again, I totally will. Because anything to make it easier to produce AI for people that donât have trillions of dollars will be extremely important.
00:34:38 Nathan Lambert: Yeah. I think that... I donât know. Iâm happy to leave it there.
00:34:43 Dean Ball: Cool.
00:34:45 Nathan Lambert: I can let you get on your trip. Itâs good to catch up. Iâm early in the process of potentially coming to DC in a few months, so I will let you know if I do.
00:34:52 Dean Ball: Oh, please do. Itâd be great to see you. We can record an episode of my podcast live.
00:34:58 Nathan Lambert: Sounds good. Okay. Thanks everybody for listening.
00:35:03 Dean Ball: Talk to yâall later. Bye.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
So-called hybrid architectures are far from new in open-weight models these days. We now have the recent Qwen 3.5 (previewed by Qwen3-Next), Kimi Linear last fall (a smaller release than their flagship Kimi K2 models), Nvidiaâs Nemotron 3 Nano (with the bigger models expecting to drop soon), IBM Granite 4, and other less notable models. This is one of those times when a research trend looks like itâs getting adopted everywhere at once (maybe the Muon optimizer too, soon?).
To tell this story, we need to go back a few years to December 2023, when Mamba and Striped Hyena were taking the world by storm â asking the question: Do we need full attention in our models? These early models fizzled out, partially for the same reasons theyâre hard today â tricky implementations, open-source tool problems, more headaches in training â but also because the models fell over a bit when scaled up. The hybrid models of the day werenât quite good enough yet.
These models are called hybrid because they mix these new recurrent neural network (RNN) modules with the traditional attention that made the transformer famous. They all work best with this mix of modules. The RNN layers keep part of the computation compressed in a hidden state to be used for the next token in the prediction â a summary of all information that came before â an idea that has an extremely long historical lineage in deep learning, e.g. back to the LSTM. This setup avoids the quadratic compute cost of attention (i.e. avoiding the incrementally expanding the KV cache per token of the attention operator), and can even assist in solving new problems.
The models listed to start this article use a mix of RNN approaches, some models (Qwen and Kimi) use a newer idea called Gated DeltaNet (GDN) and some still use Mamba layers (Granite and Nemotron). The Olmo Hybrid model weâre releasing today also falls on the GDN side, based on careful experimentation, and theory that GDN is capable of learning features that attention or Mamba layers cannot.
Introducing Olmo Hybrid and its pretraining efficiency
Olmo Hybrid is a 7B base model, with 3 experiment post-trained checkpoints released â starting with an Instruct model, with a reasoning model coming soon. It is the best open artifact for studying hybrid models, as it is almost identical to our Olmo 3 7B model from last fall, just with a change in architecture. With the model, we are releasing a paper with substantial theory on why hybrid models can be better than standard transformers. This is a long paper that Iâm still personally working through, but itâs excellent.
You can read the paper here and poke around with the checkpoints here. This is an incredible, long-term research project led by Will Merrill. He did a great job.
To understand the context of why hybrid models can be a strict upgrade on transformers, let me begin with a longer excerpt from the paperâs introduction, emphasis mine:
Past theoretical work has shown that attention and recurrence have complementary strengths (Merrill et al., 2024; Grazzi et al., 2025), so mixing them is a natural way to construct an architecture with the benefits of both primitives. We further derive novel theoretical results showing that hybrid models are even more powerful than the sum of their parts: there are formal problems related to code evaluation that neither transformers nor GDN can express on their own, but which hybrid models can represent theoretically and learn empirically. But this greater expressivity does not immediately imply that hybrid models should be better LMs: thus, we run fully controlled scaling studies comparing hybrid models vs. transformers, showing rigorously that hybrid modelsâ expressivity translates to better token efficiency, in agreement with our observations from the Olmo Hybrid pretraining run. Finally, we provide a theoretical explanation for why increasing an architectureâs expressive power should improve language model scaling rooted in the multi-task nature of the language modeling objective.
Taken together, our results suggest that hybrid models dominate transformers, both theoretically, in their balance of expressivity and parallelism, and empirically, in terms of benchmark performance and long-context abilities. We believe these findings position hybrid models for wider adoption and call on the research community to pursue further architecture research.
Essentially, we show and argue a few things:
* Hybrid models are more expressive. They can form their outputs to learn more types of functions. An intuition for why this would be good could follow: More expressive models are good with deep learning because we want to make the model class as flexible as possible and let the optimizer do the work rather than constraints on the learner. Sounds a lot like the Bitter Lesson.
* Why does expressive power help with efficiency? This is where things are more nuanced. We argue that more expressive models will have better scaling laws, following the quantization model of neural scaling.
All of this theory work is a great way to go deeper, and frankly I have a lot more to learn on it, but the crucial part is that we transition from theory to clear experiments that back it up. Particularly the scaling laws for designing this model were studied carefully to decide on the final hybrid architecture. The final performance is very sensitive to exactly which RNN block is used and in what quantity.
In scaling experiments, the results showed that for Olmo, the hybrid GDN (3:1 ratio of layers) > pure GDN (all RNN layers) > standard transformer (all attention) > hybrid Mamba2 > pure Mamba2. The crucial point was that these gaps maintained when scaling to more parameters and compute. A visual summary of the different types of architectures studied is below.
In terms of this specific model, the pretraining gains were giant! Relative to Olmo 3 dense, it represents an about 2X gain on training efficiency. When you look at evaluation performance for pretraining, there was also substantial improvement in performance, particularly after long context extension (the final 2 rows of Table 2 in the paper, highlighted below).
The journey to post-training Olmo Hybrid
Most of the experience in post-training Olmo models has been climbing up a steep curve in base model capabilities with minor tweaks to architecture. Our recipes from Tulu 2, Tulu 3, and the Olmo 3 reasoning work (building substantially on OpenThoughts 3) all worked in a fairly straightforward, off the shelf manner. Olmo Hybrid is our first experience in post-training a substantially different architecture, and the results were mixed.
1. Benchmark performance
Following the Olmo 3 recipe, we got some substantial wins (knowledge) and some substantial losses (extended reasoning) relative to the dense model. All together these still represent a very strong fully open model â just that the pretraining gains didnât translate as obviously. The results are below.
The exact reason why this happens is a research question. Our best guess is that the Olmo Hybrid base model is just a sufficiently different student model, where most of our post training data at early stages is learning from stronger âteacherâ models (a recap of this method, called distillation, appeared recently in Interconnects).
There is a lot of other research ongoing in the community around what makes a strong teacher model â generally, the best overall model is not the best teacher. In other words, training on data outputted from the model with best evaluation scores today is unlikely to unlock the ceiling in performance for your new base model. A second factor, which is even less explored, is how different base models likely need different teachers to learn from. This is why Olmo Hybrid could perform very differently, where itâs behavior is downstream of an architecture-based learning change, where the pretraining data is almost identical.
Thereâs A LOT more work to dig into here, some empirical work in generating better data and other work in understanding how different training stages fit together. I am confident this Olmo Hybrid base model is solid and more performance can be extracted, but it takes more careful work adapting existing datasets.
2. Open-source tooling
The frank reality of new architectures for open models is that the open-source software tooling support is horrific. Thereâs the paper-cuts that people are familiar with, e.g. random errors in popular libraries (as people experienced with GPT-OSS) that slow adoption, but there are also deeper problems.
A large part of the potential benefit of hybrid models is the reduction in memory usage for long-context generation, which is crucial for reinforcement learning and agentic tasks. It should be a huge win for post-training! This, unfortunately, is far from the case, and will likely take another 3-6months to get right for this batch of GDN models.
The core problem is that the open-source inference tools, e.g. VLLM, are relying on far less developed kernels (and other internals) when compared to standard transformers. This comes with two challenges â throughput slowdowns and numerical issues. Numerical issues can be combatted with a variety of inference flags. Quoting the paper again:
The two key flags in VLLM we needed to get maximum performance with the post-training model were --disable-cascade-attn, which disables cascade attention (an optimization for shared prompt prefixes), and --enforce-eager, which turns off CUDA graphs. These two flags have been used in our RL setup dating back to Olmo 3, but are new additions to evaluations. Scores for the released models drop precipitously without them. We also evaluated our final models with the hybrid model cache in the richer FP32 datatype, to improve stability via --mamba_ssm_cache_dtype following NVIDIA.
Essentially, we used these to make sure the model was numerically stable. The downside is that the inference throughput plummets, so the potential gains in compute efficiency are erased. A comparison of numbers is below.
Effectively, the 7B hybrid model today takes more compute to train with RL than our 7B dense model (that doesnât even have a common memory saving technique, GQA). The total compute estimate from the table at different context lengths is below (more visuals in the slides from my recent CMU talk).
The good news is that these are solvable problems â and improving the tooling could even improve benchmark numbers â but itâs going to take a good bit of time and hard work in the OSS community.
This leads to my final question. If Iâm optimistic about the open ecosystem evolving to support these models with ease, motivated by the better fundamental scaling of the architectures and a large cluster of leading open model builders already using it, are closed models like GPT and Claude built like this?
To be clear, this answer is a total guess (which I donât normally do), but with the evidence I have Iâd put the chance of one of the 3 frontier models being an RNN being around a coin flip. Iâll let you know if I learn for sure either way. If the scaling advantages hold at frontier scale, the economic case becomes hard to ignore, but they could already have architectures that are efficient like RNNs, but with even more benefits.
Iâm going to follow up this post with more architecture discussions, particularly on why Mixture of Expert (MoE) models are a major headache to post-train, so make sure to subscribe if that sounds interesting to you!
Thanks to Will Merrill and Finbarr Timbers for some discussions that helped inform this post.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Distillation has been one of the most frequent topics of discussion in the broader US-China and technological diffusion story for AI. Distillation is a term with many definitions â the colloquial one today is using a stronger AI modelâs outputs to teach a weaker model. The word itself is derived from a more technical and specific definition of knowledge distillation (Hinton, Vinyals, & Dean 2015), which involves a specific way of learning to match the probability distribution of a teacher model.
The distillation of today is better described generally as synthetic data. You take outputs from a stronger model, usually via an API, and you train your model to predict those. The technical form of knowledge distillation is not actually possible from API models because they donât expose the right information to the user.
Synthetic data is arguably the single most useful method that an AI researcher today uses to improve the models on a day to day basis. Yes, architecture is crucial, some data still needs exclusively human inputs, and new ideas like reinforcement learning with verifiable rewards at scale can transform the industry, but so much of the day to day life in improving models today is figuring out how to properly capture and scale up synthetic data.
To flesh out the point from the start of this piece, the argument has repeatedly been that the leading Chinese labs are using distillation for their models to steal capabilities from the best American API-based counterparts. The most prominent case to date was surrounding the release of DeepSeek R1 â where OpenAI accused DeepSeek of stealing their reasoning traces by jailbreaking the API (theyâre not exposed by default â for context, a reasoning trace is a colloquial word of art referring to the internal reasoning process, such as what open weight reasoning models expose to the user). Fear of distillation is also likely why Gemini quickly flipped from exposing the reasoning traces to users to hiding them. There was even very prominent, early reasoning research that built on Gemini!
This all leads us to todayâs news, where Anthropic named and directly accused a series of Chinese labs for elaborate distillation campaigns on their Claude models. This is a complex issue. In this post we unpack a series of questions, beginning with the impact, and ending with politics. The core question is â how much of a performance benefit do Chinese labs get from distilling from American models.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
To start, letâs review what Anthropic shared. From the blog post, emphasis mine:
We have identified industrial-scale campaigns by three AI laboratoriesâDeepSeek, Moonshot, and MiniMaxâto illicitly extract Claudeâs capabilities to improve their own models. These labs generated over 16 million exchanges with Claude through approximately 24,000 fraudulent accounts, in violation of our terms of service and regional access restrictions.
These labs used a technique called âdistillation,â which involves training a less capable model on the outputs of a stronger one. Distillation is a widely used and legitimate training method. For example, frontier AI labs routinely distill their own models to create smaller, cheaper versions for their customers. But distillation can also be used for illicit purposes: competitors can use it to acquire powerful capabilities from other labs in a fraction of the time, and at a fraction of the cost, that it would take to develop them independently.
Much like the models themselves, the benefits of distillation are very jagged. For some capabilities, particularly if you donât have a full training pipeline setup for it, quickly distilling some data from the leading frontier model in that area can yield massive performance boosts. This can definitely help the lab distilling from the API catch up much more quickly than they otherwise would. Most distillation is rather benign, using many tokens of an LLM to help process and refine existing data â putting a lot of compute into getting a few, high quality training tokens out. This sort of raw data processing work can be done on many different APIs, but one tends to be best.
When we go into what Anthropic says the three Chinese LLM builders actually used the Claude API for â as an aside, Anthropic didnât confirm that the attack was done through the API, the chat app, or Claude Code â the actual impact of the operations is very mixed. Itâs hard to know how much untracked usage these labs deployed for other projects (or other American models).
To start, Anthropic puts DeepSeek first in their blog post because theyâre the household name in the US for Chinese AI. The extent of their use is actually quite small, showing how this post is more about the big picture than the details:
DeepSeek
Scale: Over 150,000 exchanges
The operation targeted:
* Reasoning capabilities across diverse tasks
* Rubric-based grading tasks that made Claude function as a reward model for reinforcement learning
* Creating censorship-safe alternatives to policy sensitive queries
In the scale of training a language model, 150K samples is only scratching the surface as a substantive experiment. It looks like they were experimenting with some rubrics, which couldâve been for an online RL run, but thatâs extremely unlikely with how distributed the access was, and then some minor stuff on completions for sensitive queries. This usage of Anthropicâs API will have a negligible impact on DeepSeekâs long-rumored V4 model (or whichever model the data here contributed to). This was also very likely a small team at DeepSeek and unknown to much of the broader training organization.
The other two labs, Moonshot AI (makers of the Kimi models) and MiniMax reflected much broader usage.
Moonshot AI
Scale: Over 3.4 million exchanges
The operation targeted:
* Agentic reasoning and tool use
* Coding and data analysis
* Computer-use agent development
* Computer vision
MiniMax
Scale: Over 13 million exchanges
The operation targeted:
* Agentic coding
* Tool use and orchestration
The role of distillation is constantly changing. Distilling from Claude today for its agentic behavior is much more valuable than versions of Claude have been as a teacher in the past. Claude Opus 4.6 has a well-rounded agentic navigation that none of the other models quite match. Why not try training on some of the model outputs to see if your model absorbs it? Over the next few months, thatâll be less differentiated. Itâs sort of like how all the models are way better at math today than most people need â there are plenty of places to distill from.
Estimates will vary, but if each response had 10-25K tokens per exchange, the total tokens across these two labs, mostly with MiniMax, would be 150-400 billion tokens. This is a substantial amount, which could meaningfully improve a modelsâ post-training. For example, in Olmo 3 we had an SFT dataset of 20 billion tokens that could be built like this, and increasing it by 10X would be very reasonable.
These numbers are just scratching the surface of total synthetic data generation across APIs hosted by US companies. At the same time, quantity is a pretty crude way to measure impact. Just taking the outputs from Claude and figuring out how to add them to your model pipeline isnât easy. The research community has seen many cases where taking outputs from a certain teacher model unexpectedly makes the student worse â subtle interactions between the data make it variable and tricky to do this type of distillation. Itâs fundamentally a research problem.
This is what Iâm sure the Chinese labs are innovating at. Thereâs an argument that Chinese frontier labs are substantially more efficient than their Western counterparts â this is misleading.
The labs operate under different constraints. The Chinese labs are likely slightly more efficient out of necessity in being lower on resources, but overall the picture of talent access is very similar. The Chinese labs also approach benchmarks differently, making it appear that theyâre a bit closer than they really are (and appearing as if theyâre potentially surpassing). This is needed to get momentum and brand recognition in the AI market.
The Chinese labs likely innovate greatly on distilling from leading API models, due to their restricted access to GPUs. GPUs could be used to construct synthetic data, but for organizations with more funding than they can spend on research compute (being supply limited), using API-based models is one of the few other options for effectively getting more compute. Itâs way easier to figure out getting access to âbannedâ API models than it is to smuggle tens of thousands of physical GPUs and get them set up.
Itâs not only the Chinese labs that operate like this. Synthetic data from a model you donât own is all arguably distillation. Distillation is a shortcut to more compute for anyone. Itâs also a far less risky cost, as having a big cluster for research requires a very large financial commitment, where APIs are pay-as-you-go. For example, in Olmo 3 we used millions of GPU hours on the Frontier supercomputer and Azure credits through NAIRR for synthetic data. We didnât have the equivalent in GPUs (or really the cash, thank you research credits!).
All together, itâs very fair for Anthropic to be concerned about this. I still wouldnât say it is a crucial factor in these Chinese labs post-training capabilities, especially not one thatâll be easy to measure in a time gap to matching the model theyâre distilling from a la the US-China performance lag.
If we take a step back, there was even a time when Claude Sonnet was the flagship model ahead of Opus (I think this was with Sonnet 3.5), much of this comes from it being well distilled internally from Opus checkpoints. Fast iteration and high-quality data can go very far, letting student models surpass the teacher. Frontier labs use this to their advantage, by having internal-only models for generating synthetic data, but saying that Chinese models could never pass the US frontier due to data distillation is like saying that Claude Sonnet could never beat Opus. It's unlikely, and it depends a lot on release times, but with AI models making dramatic progress, weirder things like this have already literally happened.
The biggest factor unaddressed here is how distillation from stronger teacher models is harder in an era when reinforcement learning at scale is needed to train the best models. You can spend compute carefully crafting and filtering prompts, but you still need to train the model yourself with substantial, on-policy inference â generation is the majority of the compute cost for RL and it canât be generations from another model. For this reason, I expected this story to die down a bit. Itâs clear from their open research that Chinese labs have excellent RL infrastructure, despite the compute shortages.
The reason I expected it to fade is that not being allowed to distill models for âcompetitive purposesâ has violated the terms of service for API models for quite some time. Academics and open model builders in the US used to greatly worry about and debate this (and Iâve written about it multiple times in 2022 and 2023). Only later in 2024 did that worry die down in the community (and no action has been taken against any smaller model builders).
This action from Anthropic represents another continued step ratcheting up the AI geopolitical tension. Kneecapping model distillation will be far harder than restricting the shipments of physical goods like GPUs. In many ways it seems like fully restricting distillation through distributed access methods seems almost impossible, and restricting GPU sales would be far more impactful.
Anthropic and the AI industry should choose their battles. When API endpoints are available for the best models, other entities will use that to train variants of said model. This is a natural evolution of AI models. If AI models are so precious that distillation is an extreme risk, then the models will be restricted to first-party products. Anthropic has a choice to do this with their latest models. The market for API-based model alternatives may be so competitive that some companies go this path â likely in part due to Chinese models undercutting on price â but an API is a fundamental offering that no leading lab will risk walking back from anytime soon.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Last Thursday, February 5th, both OpenAI and Anthropic unveiled the next iterations of their models designed as coding assistants, GPT-5.3-Codex and Claude Opus 4.6, respectively. Ahead of this, Anthropic had a firm grasp of the mindshare as everyone collectively grappled with the new world of agents, primarily driven by a Claude Code with Opus 4.5-induced step change in performance. This post doesnât unpack how software is changing forever, Moltbook is showcasing the future, ML research is accelerating, and the many broader implications, but rather how to assess, live with, and prepare for new models. The fine margins between Opus 4.6 and Codex 5.3 will be felt in many model versions this year, with Opus ahead in this matchup on usability.
Going into these releases Iâd been using Claude Code extensively as a general computer agent, with some software engineering and a lot of data analysis, automation, etc. I had dabbled with Codex 5.2 (usually on xhigh, maximum thinking effort), but found it not to quite work for me among my broad, horizontal set of tasks.
For the last few days, Iâve been using both of the models much more evenly. I mean this as a great compliment, but Codex 5.3 feels much more Claude-like, where itâs much faster in its feedback and much more capable in a broad suite of tasks from git to data analysis (previous versions of Codex, including up to 5.2, regularly failed basic git operations like creating a fresh branch). Codex 5.3 takes a very important step towards Claudeâs territory by having better product-market fit. This is a very important move for OpenAI and between the two models, Codex 5.3 feels far more different than its predecessors.
OpenAIâs latest GPT, with this context, keeps an edge as a better coding model. Itâs hard to describe this general statement precisely, and a lot of it is based on reading othersâ work, but it seems to be a bit better at finding bugs and fixing things in codebases, such as the minimal algorithmic examples for my RLHF Book. In my experience, this is a minor edge, and the community thinks that this is most apparent in complex situations (i.e. not most vibe-coded apps).
As users become better at supervising these new agents, having the best top-end ability in software understanding and creation could become a meaningful edge for Codex 5.3, but it is not an obvious advantage today. Many of my most trusted friends in the AI space swear by Codex because it can be just this tiny bit better. I havenât been able to unlock it.
Switching from Opus 4.6 to Codex 5.3 feels like I need to babysit the model in terms of more detailed descriptions when doing somewhat mundane tasks like âclean up this branch and push the PR.â I can trust Claude to understand the context of the fix and generally get it right, where Codex can skip files, put stuff in weird places, etc.
Both of these releases feel like the companies pushing for capabilities and speed of execution in the models, but at the cost of some ease of use. Iâve found both Opus 4.6 and Codex 5.3 ignoring an instruction if I queue up multiple things to do â theyâre really best when given well-scoped, clear problems (especially Codex). Claude Codeâs harness has a terrible bug that makes subagents brick the terminal, where new messages say you must compact or clear, but compaction fails.
Despite the massive step by Codex, they still have a large gap to close to Claude on the product side. Opus 4.6 is another step in the right direction, where Claude Code feels like a great experience. Itâs approachable, it tends to work in the wide range of tasks I throw at it, and thisâll help them gain much broader adoption than Codex. If Iâm going to recommend a coding agent to an audience who has limited-to-no software experience, itâs certainly going to be Claude. At a time when agents are just emerging into general use, this is a massive advantage, both in mindshare and feedback in terms of usage data.
In the meantime, thereâs no cut-and-dried guideline on which agent you need to use for any use-case, you need to use multiple models all the time and keep up with the skill that is managing agents.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
Assessing models in 2026
There have been many hints through 2025 that we were heading toward an AI world where benchmarks associated with model releases no longer convey meaningful signal to users. Back in the time of the GPT-4 or Gemini 2.5 Pro releases, the benchmark deltas could be easily felt within the chatbot form factor of the day â models were more reliable, could do more tasks, etc. This continued through models like OpenAIâs o3. During this phase of AIâs buildout, roughly from 2023 to 2025, we were assembling the core functionality of modern language models: tool-use, extended reasoning, basic scaling, etc. The gains were obvious.
It should be clear with the releases of both Opus 4.6 and Codex 5.3 that benchmark-based release reactions barely matter. For this release, I barely looked at the evaluation scores. I saw that Opus 4.6 had a bit better search scores and Codex 5.3 used far fewer tokens per answer, but neither of these were going to make me sure they were much better models.
Each of the AI laboratories, and the media ecosystems covering them, have been on this transition away from standard evaluations at their own pace. The most telling example is the Gemini 3 Pro release in November of 2025. The collective vibe was Google is back in the lead. Kevin Roose, self-proclaimed âAGI-pilledâ NYTimes reporter in SF said:
There's sort of this feeling that Google, which kind of struggled in AI for a couple of years there â they had the launch of Bard and the first versions of Gemini, which had some issues â and I think they were seen as sort of catching up to the state of the art. And now the question is: is this them taking their crown back?
We donât need to dwell on the depths of Geminiâs current crisis, but they have effectively no impact at the frontier of coding agents, which as an area feels the most likely for dramatic strides in performance â dare I say, even many commonly accepted definitions of AGI that center around the notion of a âremote worker?â The timeline has left them behind 2 months after their coronation, showing Gemini 3 was hailed as a false king.
On the other end of the spectrum is Anthropic. With Anthropicâs release of Claude 4 in May of 2025, I was skeptical of their bet on code â I was distracted by the glitz of OpenAI and Gemini trading blows with announcements like models achieving IMO Gold medals in mathematics or other evaluation breakthroughs.
Anthropic deserves serious credit for the focus of its vision. They were likely not the only AI lab to note the coming role of agents, but they were by far the first to shift their messaging and prioritization towards this. In my post in June of 2025, a month after Claude 4 was released, I was coming around to them being right to deprioritize standard benchmarks:
This is a different path for the industry and will take a different form of messaging than weâre used to. More releases are going to look like Anthropicâs Claude 4, where the benchmark gains are minor and the real world gains are a big step. There are plenty of more implications for policy, evaluation, and transparency that come with this. It is going to take much more nuance to understand if the pace of progress is continuing, especially as critics of AI are going to seize the opportunity of evaluations flatlining to say that AI is no longer working.
This leaves me reflecting on the role of Interconnectsâ model reviews in 2026. 2025 was characterized by many dramatic, day-of model release blog posts, with the entry of many new Chinese open model builders, OpenAIâs first open language model since GPT-2, and of course the infinitely hyped GPT-5. These timely release posts still have great value â they center the conversation around the current snapshot of a company vis-a-vis the broader industry, but if models remain similar, theyâll do little to disentangle the complexity in mapping the current frontier of AI.
In order to serve my role as an independent voice tracking the frontier models, I need to keep providing regular updates on how Iâm using models, why, and why not. Over time, the industry is going to develop better ways of articulating the differences in agentic models. For the next few months, maybe even years, I expect the pace of progress to be so fast and uneven in agentic capabilities, that consistent testing and clear articulation will be the only way to monitor it.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
One of the big stories of 2025 for me was how Nvidia massively stepped up their open model program â more releases, higher quality models, joining a small handful of companies releasing datasets, etc. In this interview, I sat down with one of the 3 VPâs leading the effort of 500+ technical staff, Bryan Catanzaro, to discuss:
* Their very impressive Nemotron 3 Nano model released in Dec. 2025, and the bigger Super and Ultra variants coming soon,
* Why Nvidiaâs business clearly benefits from them building open models,
* How the Nemotron team culture was crafted in pursuit of better models,
* Megatron-LM and the current state of open-source training software,
* Career reflections and paths into AI research,
* And other topics.
The biggest takeaway I had from this interview is how Nvidia understands their unique roll as a company that and both build and directly capture the value they get from building open language models, giving them a uniquely sustainable advantage.
Bryan has a beautiful analogy for open models this early in AIâs development, and how they are a process of creating âpotential energyâ for AIâs future applications.
I hope you enjoy it!
Guest: Bryan Catanzaro, VP Applied Deep Learning Research (ADLR), NVIDIA. X: @ctnzr, LinkedIn, Google Scholar.
Listen on Apple Podcasts, Spotify, YouTube, and where ever you get your podcasts. For other Interconnects interviews, go here.
Nemotron Model Timeline
2019â2022 â Foundational Work
* Megatron-LM (model parallelism framework that has become very popular again recently; alternatives: DeepSpeed, PyTorch FSDP).
* NeMo Framework (NVIDIAâs end-to-end LLM stack: training recipes, data pipelines, evaluation, deployment).
Nov 2023 â Nemotron-3 8B: Enterprise-ready NeMo models. Models: base, chat-sft, chat-rlhf, collection. Blog.
Feb 2024 â Nemotron-4 15B: Multilingual LLM trained to 8T tokens. Paper.
Jun 2024 â Nemotron-4 340B: Major open release detailing their synthetic data pipeline. Paper, blog. Models: Instruct, Reward.
JulâSep 2024 â Minitron / Nemotron-Mini: First of their pruned models, pruned from 15B. Minitron-4B (base model), Nemotron-Mini-4B-Instruct. Paper, code.
Oct 2024 â Llama-3.1-Nemotron-70B: Strong post-training on Llama 3.1 70B. Model, collection. Key dataset â HelpSteer2, paper.
MarâJun 2025 â Nemotron-H: First hybrid Mamba-Transformer models for inference efficiency. Paper, research page, blog. Models: 8B, 47B, 4B-128K.
May 2025 â Llama-Nemotron: Efficient reasoning models built ontop of Llama (still!). Paper.
Sep 2025 â Nemotron Nano 2: 9B hybrid for reasoning, continuing to improve in performance. 12B base on 20T tokens (FP8 training) pruned to 9B for post-training. Report, V2 collection.
Nov 2025 â Nemotron Nano V2 VL: 12B VLM. Report.
Dec 2025 â Nemotron 3: Nano/Super/Ultra family, hybrid MoE, up to 1M context. Super/Ultra H1 2026. Nano: 25T tokens, 31.6B total / ~3.2B active, releases recipes + code + datasets. Papers: White Paper, Technical Report. Models: Nano-30B-BF16, Base, FP8.
Nemotronâs Recent Datasets
NVIDIA began releasing substantially more data in 2025, including pretraining datasets â making them one of few organizations releasing high-quality pretraining data at scale (which comes with non-negligible legal risk).
Pretraining Data
Collection â CC-v2, CC-v2.1, CC-Code-v1, Code-v2, Specialized-v1, CC-Math-v1. Math paper: arXiv:2508.15096.
Post-Training Data
Core post-training dumps (SFT/RL blends):
* Llama Nemotron Post-Training v1.1 (Apr 2025)
* Nemotron Post-Training v1 (Jul 2025)
* Nemotron Post-Training v2 (Aug 2025)
2025 reasoning/code SFT corpora:
* OpenMathReasoning (Apr 2025)
* OpenCodeReasoning (Apr 2025), OpenCodeReasoning-2 (May 2025)
* AceReason-1.1-SFT (Jun 2025)
* Nemotron-Math-HumanReasoning (Jun 2025), Nemotron-PrismMath (Apr 2025)
NeMo Gym RLVR datasets: Collection
Nemotron v3 post-training (Dec 2025): Collection
HelpSteer (human feedback/preference):
* HelpSteer (Nov 2023)
* HelpSteer2 (Jun 2024)
* HelpSteer3 (Mar 2025)
And others, not linked here.
Chapters
* 00:00:00 Intro & Why NVIDIA Releases Open Models
* 00:05:17 Nemotronâs two jobs: systems R&D + ecosystem support
* 00:15:23 Releasing datasets, not just models
* 00:22:25 Organizing 500+ people with âinvitation, not controlâ
* 0:37:29 Scaling Nemotron & The Evolution of Megatron
* 00:48:26 Career Reflections: From SVMs to DLSS
* 00:54:12 Lessons from the Baidu Silicon Valley AI Lab
* 00:57:25 Building an Applied Research Lab with Jensen Huang
* 01:00:44 Advice for Researchers & Predictions for 2026
Transcript
00:00:06 Nathan Lambert: Okay. Hey, Bryan. Iâm very excited to talk about Nemotron. I think low-key, one of the biggest evolving stories in twenty-five of open models, outside the obvious things in China that everybody talks about, that gets a ton of attention. So th- thanks for coming on the pod.
00:00:22 Bryan Catanzaro: Oh, yeah, itâs my honor.
00:00:23 Nathan Lambert: So I wanted to start, and some of these questions are honestly fulfilling my curiosity as a fan. As like, why does NVIDIA, at a basic level, release Nemotron as open models?
00:00:39 Bryan Catanzaro: Well, we know that itâs an opportunity for NVIDIA to grow our market whenever AI grows, and we know that having access to open AI models is really important for a lot of developers and researchers that are trying to push AI forward. you know, we were really excited by efforts from some other companies around the industry to push openly developed AI forward. You know, Meta did some amazing work, obviously, with Llama and you know OpenAI released GPT OSS, which was exciting. And the Allen Institute, of course, has been, you know, really leading the charge for research, open research and, you know, also things like the Marin Project and OpenAthena. You know, like thereâs, thereâs a bunch of things that weâre always excited to see develop.
And, you know, as we think about where AI is gonna go, you know, NVIDIA believes that AI is a form of infrastructure. itâs.. AI is a very useful technology when itâs applied, but on its own you know, itâs kind of a foundation and infrastructure. We think that technology generally works better when thereâs openness to the infrastructure so that people can build things in different ways. You know, you think about the way that the internet transformed every aspect of the world economy is pretty profound, and weâre not done yet.
But the way that, for example, retail uses the internet is different from the way that healthcare uses the internet. And the fact that you know, different sectors of the economy were able to figure out how to incorporate the internet into the beating heart of their businesses in different ways was possible because the internet was built on open technologies that, you know, allowed people to try different things. And we think AI is gonna evolve in a similar way, that organizations across every sector of the world economy are gonna find new and surprising and fun, and important things to do with AI, and theyâll be able to do that better if they have the ability to customize AI and incorporate it directly into the work that they do. and so -- and by the way, this is not to detract from any of the you know, more closed approaches to AI, you know, the APIs that we see from a number of leading labs that, you know, are just extraordinary and have amazing capabilities. Weâre excited about those, too.
You know, NVIDIA loves to support AI in all of its manifestations, but we feel like right now the sort of closed approaches to deploying AI are doing pretty well but we, you know, could use some more energy in the openly developed AI ecosystem, and so thatâs why weâve been putting more effort into it this past year.
00:03:42 Nathan Lambert: Yeah. So Iâm definitely gonna dig into this a lot âcause I have seen this. Weâre sitting here recording in January twenty-six, which is in the midst of the rollout of these Nemotron three models. Thereâs the-- I think the Nano has released in the fall, which was probably one of the biggest splashes the org has made, and everybodyâs eagerly awaiting these super and ultra-larger variants.
And itâs like how far are you, how far are you willing to push this Nemotron platform? Like, is it just depending on the users and the uptake and the ecosystem? Like, like, what is the-- is there a North Star in this? Or you hear a lot of.. if you listen to a lot of other open labs, theyâre like: âWe want to build open AGI,â which is like, I donât necessarily think grounded, but thereâs like a very unifying vision.
Is there something that you try to set the tone for it that goes through the organization? I mean, AI too, itâs like-
00:04:31 Bryan Catanzaro: You know, my North-
00:04:32 Nathan Lambert: .. academics is so-
00:04:34 Bryan Catanzaro: For Nemotron.
00:04:36 Nathan Lambert: Okay, go ahead.
00:04:37 Bryan Catanzaro: Oh, sorry. Go ahead.
00:04:39 Nathan Lambert: I was just, like, gonna compare to, like, AI too, where we can have such a-- like, we have a very specific vision, being so open that itâs like, I think, like, research is so needed, and thereâs so little recipes to build on, like, with really credible research. So thereâs, like, a research infrastructure, and then when you have something like Llama, it was, like, built on Zuckerbergâs vision, and he changed his mind, which I actually thought his vision was ex- was excellent, the way he articulated the need for open models, and it kind of faded. So itâs like, is there a way to set a vision for an org that, like, permeates every- everyone and is really compelling and exciting?
00:05:17 Bryan Catanzaro: Right. Well, we built Nemotron for two main reasons. The first is because we need to for our main product line. So what I mean by that?
Well, accelerated computing, what NVIDIA does, we build fast computers, right? But the point of building fast computers is to help people do new things. and actually every fast computer is also a slow computer. you know, the observation that it would be nice if computers were faster and could do more things isnât new. thatâs been around since the beginning of computing. So what makes accelerated computing different from standard computing is that weâre prioritizing, you know, weâre focusing, weâre deciding weâre gonna accelerate this workload. This other workload, which is like ninety-nine percent of all of the workloads, weâre gonna let somebody else do that, right?
So, like, you do not buy NVIDIA systems to do any general purpose computation. You buy them for a purpose, right? Which is these days, all about AI. But when you think about the workload, the compute workloads involved in AI thereâs a, thereâs a lot of diversity and thereâs a lot of really important -.. parameters, hyperparameters, or algorithmic approaches that all have enormous imp- impacts on the systems that we need to build for AI.
So things like numeric precision MoE architecture, which of course, influence net-- it influences network design. you know, weâre dreaming about sparsity. We, you know, weâve had, weâve had sparse neural network acceleration in the GPU since Ampere. I donât think that itâs being used enough. you know, so how do we, how do we figure out how to use that? These, these sorts of things have an enormous impact on the future of NVIDIAâs main product line, and we have to understand the answers to those questions deeply ourselves in order to know what weâre going to build.
We canât just go to our customers and do a survey and say, âHey â you know, Meta, for example, since we were just talking about them, âwhat would you like to see in a future product line from NVIDIA?â Of course, Metaâs always trying to help us as much as they can, but thereâs limits to what they can tell us because, you know a lot of the information that influences the design of these systems, itâs very expensive to derive, and so therefore, itâs, itâs very closely held. And so we need to be able to understand these questions very deeply in order to understand what kind of systems to build, in order to understand what weâre accelerating in AI and what weâre not gonna worry about. and so thatâs kind of the first job for Nemotron models, is to make it possible for NVIDIA to continue to exist as a company. And I think itâs important that the community knows that because thatâs the reason why NVIDIA is making the investments in Nemotron, is because we believe itâs essential for the future of our company. and so this isnât-- and although as much, as much as it feels good to say, you know, NVIDIA believes in open openly developed AI because you know, weâre so charitable, but actually, thatâs not the case. This is actually a business decision-
00:08:34 Nathan Lambert: Itâs smart
00:08:34 Bryan Catanzaro: .. like, for NVIDIA, our business needs us to know about AI very deeply. And and so, you know, the amount of investment that is justified to carry on NVIDIAâs ongoing business, I think, is large. and so thatâs thatâs job number one for Nemotron. Now job number two for Nemotron is to support the ecosystem more broadly outside of NVIDIA. and, you know, NVIDIA has a special position in the AI landscape. of all of the big AI companies I think weâre the one that works with the most other companies. We support every company small and large, AI native company to old established enterprise.
We work with hyperscalers, we work with tiny little startups, we work with countries around the world. so we have this unique position and I think also a uni- unique responsibility and al- maybe also a unique opportunity, that whenever AI is able to grow in any sort of direction, in any capability, then you know, thatâs an opportunity for us to grow our business. Obviously, itâs not automatic, right? you know, the AI market is diverse, and itâs getting more diverse, and it should be, âcause itâs the most important market in the history of humanity. So so we acknowledge that, and at the same time, we know that itâs in our interest to develop the AI ecosystem. The more people that are building, inventing, and deploying AI, the more opportunity that we have as a company.
So thatâs job number two for Nemotron.
00:10:17 Nathan Lambert: Yeah. I really appreciate you saying it so directly âcause itâs like weâve worked.. We- I launched this thing, the Adam Project, last summer, which is trying to get more investment in the US open models, and itâs like the only company that has an obvious business model for open models is something like NVIDIA, where you need to make sure that the open models and the research ecosystem plays nicely on CUDA, because then youâre gonna be able to be one-- Youâre so many steps closer to research thatâs happening. If not, like, if it like- Thereâs such an advantage to have research happen mostly on GPUs relative to AMD or anything like this, so.
00:10:49 Bryan Catanzaro: Well, you know, we are-- weâre, weâre not thinking about how to prevent competition. You know, we welcome competition. Thereâs lots of competition. There should be more competition in this space, but we are very self-interested in staying engaged with the community.
You know, itâs very important. You know, CUDA not many people remember this because it happened so long ago, but you know, CUDA started out with a lot of outreach from NVIDIA to the academic and industrial community saying, âHey, we have this new way of doing computing. weâd love to see what you can do with it.â In fact, you know, I started using CUDA in 2006 when I was a grad student at Berkeley because David Kirk, who was the chief scientist of NVIDIA at the time, came over to Berkeley and said, âHey we just released this new GPU, and it has this new programming model called CUDA. You should give it a try.â And I was-- at the time, I was working on machine learning on FPGAs, and I had been working on this one particular piece of support vector machine training on the FPGA, and I decided to take that little piece and write it in CUDA, and it took me like fifteen minutes, and then I ran it, and it was like two hundred times faster than my single-threaded CPU code, and I was like: âWhoa, that was way easier than what I was doing before. Iâm just gonna go do that,â right?
So, like, my own personal involvement with CUDA and NVIDIA came about because of this outreach that NVIDIA conducted right from the beginning of CUDA. you know, of course, that led to a lot of great things for NVIDIA, including AlexNet, which was another academic project, you know, where Alex Krizhevsky and Ilya Sutskever were thinking about: âHow do we train larger neural networks on more data? weâre gonna go write a bunch of GPU code that uses the GPU in a, in a kinda new and clever way, so that we can train a better image classification model.â And, you know, that had such astonishing results, it kicked off the deep learning era for the whole community. and again, not something that-.. could have been done top-down. That was a, that was a very much a result of NVIDIA supporting open development and re- research in parallel computing and artificial intelligence. And so we remember that, and weâre thinking about in twenty-six, what does it look like to help, you know, the Alex Krizhevsky of the future, whoâs, whoâs a grad student in a lab somewhere, invent the next technology that changes the world? It seems really difficult to do that without something like Nemotron or, or the other openly developed AI projects out there. yeah, I also wanna say in regards to this Nemotron is not trying to be the only project out there.
Weâre part of the community. We love other people doing great work in openly developed AI. We learn from things that other people do and you know, so weâre, weâre trying to support the community because itâs in our interest, but we you know, weâre very happy to see other people contributing as well.
00:13:57 Nathan Lambert: Yeah, I mean, I can transition into something I wanted to ask about is like, I see multiple ways, twenty-five Nemotron mat-- in, I donât wanna use the word maturing âcause I wanna ask you about how it feels in the org, but just like the output reached levels that were more noticed by the community and people building with models. And thereâs a lot of ways that can happen, but one of them is like, in my niche community, Iâve been using Nemotron datasets a lot. Like we-- when we redo our post-training recipe, one of the only people we look at is like, okay, NVIDIA, Nemotron has released a lot of high-quality, openly licensed post-training data. this year, you also started releasing some pre-training data, which among AI2 got a lot of notice. Like, what is that? is that like a distinct shift within Nemotron?
Is that something that youâve wanted to do for a while and finally just did? But itâs âcause itâs like-- it is just like a zero to one moment where releasing pre-training data comes with legal risk for any company, but so few people do it, where on my side of the world, itâs like pretty easy to normally say what the best pre-training dataset is, and it had, for a long time, oscillated between like Hugging Face, AI2, DCLM, and there was like literally only two or three options. So in terms of fundamental research, like I think thatâs a big step from an org to support the community and take on some risk. So if you have any story you can tell and or just say like, I appreciate it, thatâs, thatâs all.. thatâs all I got.
00:15:23 Bryan Catanzaro: Well, yeah. I mean, so I think itâd be great if more people could understand that Nemotron is not just a model, right? Like, what weâre trying to do with Nemotron is to support openly developed AI, because, again, thatâs our big opportunity, right? Now, thereâs a lot of organizations that are incentivized to build a model, and the model is maybe the thing that runs their business, right?
But at NVIDIA, the model is not the thing that runs our business, itâs the systems. So when weâre thinking about how do we support the ecosystem, itâs clear to us that the ecosystem needs more than just a model. Thereâs a lot of models out there already, you know? And of course, we want Nemotron to be awesome, but you know, if Nemotron can convince other people to work on AI because of a dataset or a technique, you know, weâre, weâre trying to be very open with all of the things we learn, you know, including..
I mean, we do a lot of expensive experiments in order to figure out how to do blending for our datasets or to figure out, you know, optimize our settings and, you know, these sorts of things. weâre very happy for other people to pick that up and run with it if itâs useful to them, you know. And so that makes Nemotron a different kind of AI effort. Of course, there is a model component, and thatâs a tangible thing, and itâs, itâs easy to focus on that, but we see Nemotron as you know, an effort that includes models, but also includes datasets, techniques, all of all of the research that goes into Nemotron. And again weâre a unique kind of AI organization because of the way that we work with AI companies around the industry and because of the way that our business works, we can afford to be more open with some of these things than maybe some other organizations could be.
Now to your question about, like, does it take some courage in order to be open? Yeah, absolutely it does. and you know, I think thereâs been-- one of the things thatâs happened in twenty-five is that thereâs been an evolving understanding within NVIDIA about the benefits of openness, and that has really enabled the company to make some investments that perhaps it was a little gun-shy to make in the past. And so thatâs really encouraging for me. itâs something that Iâve you know, advocated for a while, and so itâs, itâs great to see the company kind of lining up behind it. I also, you know, to your point about like twenty-five being a, a year where Nemotron really made some strides, I want to say thank you for noticing that, and then maybe tell you a little bit about how that happened, because I think itâs instructive for me about how I think the work is gonna go forward in the future.
So you know, NVIDIA is a very decentralized company with a lot of volunteers. You know, everybody that works at NVIDIA is a volunteer. And what do I mean by that? Well, I mean, look, the industry is moving quick.
You know, people can always move from one job to the next. So the way that we think about the work that we do is like, itâs very decentralized, itâs very much let smart people figure out what they should be doing and then kind of self-organize. Now one of the challenges of self-organization in a field thatâs moving quickly is that sometimes a whole bunch of people decide to-.. do similar kind of overlapping things but arenât really coordinated. and thatâs okay at the beginning because, you know in a place like NVIDIA, itâs just great to have some energy. It, it took us a while, I think, as a company to figure out that Nemotron was better together.
That rather than having, like, this group has a, has a model and that group has a dataset, and like, you know, then we end up publishing papers that kind of you know donât really acknowledge each other and arenât really coordinated. And then, of course along with that, we need to have k times the GPUs, where k is the number of independent efforts. we realized that, you know building AI, you really do need to figure out how to collaborate. the AI efforts that are built from teams of people focused on the overall effort succeeding rather than their own particular piece of the project succeeding, those are the ones that, you know, really change the world. And, you know, of course, NVIDIA works that way for the systems that we build, right? So, like, the people working on the memory controller on the GPU know that they also have to work with the people working on the SM that does the math, right?
Like, you canât, you canât make a GPU where itâs just like, âWell, weâve got an awesome memory controller,â if the math doesnât work, right? It all has to, has to kinda work together. And so that coordination, I think in the field of AI, it took us a little bit longer to do maybe than you could imagine that it could have. and I think that slowed the progress for Nemotron. so I give a lot of credit to the Nemotron team for realizing over the past, I donât know, year and a half or so, that it was really time to join up and build one thing and make it awesome, and deeply understand that the success of the Nemotron project was more important than the success of any individual piece of that project. And the reason why Iâm telling you all of this is because I think thatâs actually true more broadly than just inside NVIDIA, and I think itâs, itâs difficult. you know, researchers like those of us with PhDs, for example, we are taught how to be independent, you know, and how to, how to build up our Google Scholar profile, and thereâs, like, an incentive to go ahead and focus on that.
And a lot of successful academics and people researchers you know, they manage to push that pretty far and get some pretty amazing results. But, you know, I do believe that in 2020- in the 2020s you know, that the best research is done as part of a larger team. so how do we figure out how to work together? You know, how do we figure out how to put the success of the team first? That is a thing that is challenging to do but if we can achieve it, I think yield significant results.
And, you know, to the extent that we made progress in that part of the organization, I think we also saw progress in the technology. and thatâs.. That gives me great hope for 2026 for Nemotron because the way the team is working together, I think is you know, pretty extraordinary. Thereâs just an enormous number of brilliant people that have decided that theyâre gonna volunteer to make Nemotron awesome, and weâre, weâre starting to see some pretty great things come together.
00:22:25 Nathan Lambert: I agree with everything you said. Do you have any advice for making the orgs come together? I think weâve seen big-- Wait, Iâve seen two class-- thereâs two classes of AI companies right now. One is startup, does everything, and you have a model in six months, but youâre building from zero, and you have-- you p-- everybody agrees when they start that they do this. And then you have Googleâs famous long-winded reorgs, which they actually eventually got right. Like, they got it very right with whatâs going on with Gemini and Google DeepMind-.. right now. And itâs like, do you have any advice on doing this? I think, like, Iâm, AI too, also advocating for this, but itâs very hard. I think personally-
00:22:58 Bryan Catanzaro: Itâs-
00:22:58 Nathan Lambert: .. itâs like, I mean, Iâm, Iâm a special case âcause Iâm also visible, where itâs e-- very easy for me to turn internet activity into, like, reputation points because of algorithms and size. But itâs very hard to do bottom-up technical work and get all of this and get all the culture alignment. So do you have any advice on actually, like, what works in this domain?
00:23:20 Bryan Catanzaro: You know whatâs worked for us is invitation and not control. so you know, one way that, like, for a while I kinda wanted to try to implement was, like, nobody gets to publish any papers in AI unless theyâre clearly part of Nemotron. So this is kind of a top-down, like, weâre gonna make you do it, right? I came to the realization that which we never implemented this, by the way, but I came to realization that this was a bad idea because it would just breed resentment, and, you know, NVIDIA is a company of volunteers. Everybody here is a volunteer.
So what we need to do is create the conditions by which it makes sense for people to volunteer to be part of Nemotron. And so the way that we went about doing that first of all it involved like, some top-level agreements between me and some of the other leaders of Nemotron, for example, John Cohen and Kerry Briski. I work very closely with the two of them. And you know, that hadnât always been the case.
Like, we kind of had all come to this place independently. but we realized, like, Nemotron, better together, all three of us, and then we started telling our teams that: âYou know, we really think Nemotron is gonna be better together.â so that top-down alignment, I think was really helpful. We-- again, we werenât telling people exactly what to do, but we were just sending a con constant message like, you know, âNemotronâs better together.â And then we built some structures that facilitated collaboration. So in the past decisions in the Nemotron project tended to be made in kind of a an opaque way. and the reason for that is just, you know-.. itâs hard to tell everybody about the middle of the sausage-making process. You know, itâs, like, messy and dif- difficult, and so, like, you know, itâs natural.
Like, researchers, weâre used to doing this, right? Itâs a fait accompli. Like, âHereâs my ICML paper,â and like, you know, the fact that you spent, like, two years failing at that task before you finally succeeded, and then you tied a bow around it and gave it to the ICML committee, you donât really talk about that, right? And so itâs difficult for researchers to, to be open about the middle of the process of research.
Thereâs a lot of failure, and itâs hard for people to feel like theyâre, theyâre not looking amazing. But what we, what we decided to do is we structured the project with.. Thereâs about twenty different areas for the project. Each of them has a clear leader, what we call a pilot in command.
Their job is to-- the job of the pilot in command is to land the airplane. You know, you just want the airplane to land, okay? So somebody, if youâre landing an airplane, there might be multiple pilots on board, but only one of them is gonna land the airplane at any time, right? Because it would be chaos if two of them tried to land at the same time, people would die.
So so this is not a committee structure; it is a delineated responsibility structure. And then the purpose of that pilot in command for each of these sections is to gather together all the best ideas, help the group of people that are interested in working on that space to come up with data-driven answers to what we should do, what technical decisions we should make, and then document that, you know, in a, in a way that other people can review. and you know, the thing thatâs been really great about that is that it is inviting to people because when they see, like, okay, hereâs the group of volunteers that are working on this area of Nemotron and then they want to contribute, itâs much clearer about how they could go about doing that, and itâs also clearer what the group needs because you know, these meetings are being held in the open. and we have-- we actually have a website where all of the ideas are submitted. they each get, like, a unique identifier, and then they get engaged with, you know, the PIC is trying to understand what the implications are, what kinds of experiments need to be run in order to prove or disprove the idea? how do we do what I call integration studies? You know, I, integration studies are so key for bringing researchers together, and theyâre so opposite of what we are taught when weâre learning how to do ablations as a graduate student. You know, rather than, like, isolating the particular contribution of one idea, integration studies are about putting a hundred ideas together and seeing if theyâre better than what we had before. so this kind of thing, doing that in a structured way and in a, in an open way internally has then made it possible for more people to volunteer, and that has then generally raised the rigor of the experiments and also the I think the outcome of the work.
00:28:15 Nathan Lambert: Yeah, this is great. I think that over the last few years, thereâs been more consensus on things that work for research. And I think the- we also do integration tests very regularly of like, is this feature gonna land for the model? And thatâs kind of a..
Itâs a good- itâs a nice mirror to ablations, where we know research is changing so much. Thereâs a lot of turmoil in the academic research community, and itâs nice to have things that are tangible as ways that are a little bit different when youâre doing these large-scale projects. So people that underst- like, you still need to do ablations. But then it needs to survive, like, an additional test in order to land into the model.
So itâs like an additional type of work that needs to be done, and I just like to have words to describe what is actually happening. I think on the Nemotron-3 Nano front, I do a lot of analysis on just looking at basic adoption metrics and Nemotron we created this, what we called like a relative adoption metric, which is essentially looking at downloads over time for models, because itâs easy to know which models have a ton of downloads that are released a while ago. But to, like, look at the trajectory of downloads changing over time, this is a lot-- this is a mouthful. Itâs kind of an aside, but, like, Nemotron Nano 3 was in the thirty B size range, like, on track to be one of the top ten models downloaded of all time.
The point that I bring this up, other than to just flatter you, is like, do you think last mile adoption takes a substantial amount of work other than making, like, a very functional model? Or does adoption-- like, do you need to, like, change the recipe that youâre making and put a lot of focus and evaluation and, like, change this over time so that you actually get people to really use the model, rather than, like, âOh, the benchmarks are good,â look at NVIDIA flying high?
00:30:03 Bryan Catanzaro: Right. Yeah, I mean, wow, it has taken the whole company coming together in order to make Nano V3 have more of an impact than the models that we released before. and thereâs so many different aspects to that. obviously, thereâs a lot of technical aspects which frankly, I think we have more work to do. So, like you know, making sure that on day zero, when we release something, that the quantizations, all the quantizations, the best quantizations are out there, that the speed on all of the important inference frameworks is out there, that it runs on all of the edge devices that we care about fla- flawlessly, that the install experience is great. You know, this kind of work is extraordinarily important because you know, itâs a crowded world.
Thereâs so many different things that people could choose to work with, and any amount of friction that gets in the way of people even evaluating something that you do is gonna blunt the results, no matter how good that technology is.. I donât think that weâre amazing at this yet, so this is something that I anticipate weâre gonna see a lot more investment in as the, you know more people at NVIDIA from all over the company, from marketing, from developer relations, from software engineering, you know as they-- as we all come together in support of this effort. so yeah, so it does, it does take an enormous amount of work. and then, you know, something that Iâm particularly interested in is you know, how do we work engage-- i-in a new way, sort of engage with the community to make future Nemotron models even stronger? You know if the only things that we were to optimize for with a Nemotron model would be kind of academic benchmarks that are, you know, highly cited itâs likely the case that the model wouldnât be general enough to really be useful. And so what weâre trying to build is a technology that other people can extend and deploy, and that means we need to have, like, other ways of understanding the strength of a model besides you know, a handful of academic benchmarks.
I think we have a lot of room to grow here. Iâm hoping over time that we develop the muscle of being able to engage with the community and learn from them. Like, you know, okay, this particular thing that I tried to do with Nemotron, it didnât work. It did this other thing that, you know, I wasnât expecting, it was wrong. well, that can become feedback that then is used to make the next version better.
I think weâve got a lot of work to do in that regard.
00:33:10 Nathan Lambert: Do you think thereâs any magic to it? Iâve-- Iâm blown away by how successful OpenAIâs two open-source models are. Like, yes, theyâre obviously the number one name brand in AI, but on the same metric that I see you guys, like, overperforming, like, what I would expect. Iâm like, âWow, great job, NVIDIA.â Theyâre, like, totally off the charts, like, on track to like, beat Llamaâs, like, most downloaded numbers ever with these two GPT OSS models.
And I feel like what they-- like, even on release, they had hiccups where people were pretty negative on it. But for whatever reason, it has just like.. People figured it out, and it just clicked, and then just, like, for a company to say so little about it. Like, we-- Meta put so much effort into Llama being adopted, and you obviously are putting a lot of effort into this.
Like, Iâm just like, did OpenAI just crack the code, or is there sometimes a bit of luck?
00:33:59 Bryan Catanzaro: Well, I donât think I, I donât think about OpenAI as a, as a lucky company. I think of them as a visionary company that works incredibly hard and you know, I think their success is well deserved. I love the GPT OSS models. You know definitely theyâre an inspiration for us here at Nemotron. and yeah, so I think OpenAI also has, like, some other ways of engaging with the community just because of the large number of people that use their services, and that helps them learn things about what are people trying to do with AI, that then they can address when theyâre building models, and you know, obviously, you know, people talk about that as a flywheel. you know, I think thatâs really interesting and really important.
NVIDIA is never going to have the same kind of flywheel as OpenAI does. Weâre not trying to build a service like ChatGPT. What weâre trying to do is help the ecosystem, you know, be strong and enduring. we think that itâs important for there to be this openly developed AI ecosystem, and also weâre, weâre trying to build our next generation of systems, and so we have our own reasons for doing this. But weâre not ever going to have the same exact user base or flywheel that OpenAI does.
On the other hand, you know, we are able to work with institutions around the world in our own way, that I think offers us different opportunities and hopefully, that helps us make things that are, that are useful, too.
00:35:38 Nathan Lambert: Yeah, this makes me realize, Iâm having a lot of conversations on.. There are many open model efforts, especially even among people that are fully open, and itâs like, how do we better coordinate? So especially at the smaller scale, itâs like AI2 and Hugging Face. So theyâre not big teams.
Like, how do we make sure weâre not doing the same data project at the same-- the same exact thing at the same time? And itâs like, I wonder if thereâs opportunities for open companies, like LM Arena has historically released a lot of user data to, like, better help us close this kind of what are people using models for flywheel. And but itâs just-- itâs very hard to build cross-organizational model improvement pipelines, is something that I think. I think models become pretty vertical in terms of somebody at NVIDIA getting the feedback and the model making better.
So thatâs what would be something I would like to see this year, but I donât have ideas for doing it well.
00:36:28 Bryan Catanzaro: Yeah. You know at NVIDIA, we have a tradition of working really closely with, you know, organizations that use our technology. and, you know, we really-- we have, we have teams of engineers that their job is to enable success for our customers. in fact, thereâs more people at NVIDIA that care about the success of people outside of NVIDIA than I feel like sometimes there are people that care about the success of things inside NVIDIA. So, like, sometimes Iâm like, Iâm like: âHey, could we use a little bit of that e-energy to support Nemotron?â And, and the answer is yes, and NVIDIA is doing that. But I think as Nemotron matures, weâre gonna find that you know, the organizations that work with NVIDIA to make Nemotron awesome for their business, for their use case are gonna have a say in how Nemotron evolves and hopefully, that helps Nemotron address their needs.
00:37:29 Nathan Lambert: .. Yeah, a basic question: how many people, like, how many employees does it take to build all the different versions of Nemotron? I havenât brought this up because you also have other great types of models. I think our, like, open model analyst, Florian, is obsessed with the Parakeet model, âcause- Much faster at typing and is much faster at speaking than typing.
So thereâs a lot of other-- I donât know-- I donât have the full list of other NVIDIA models off the top of my head, but you are releasing a lot of varieties of models. So I think itâs a bit of a thereâs more context to my original question, which is I think about language models âcause Iâm a n-- like, I just think of AIâs progress is gonna continue to go very fast, so I focus as that as the engine. So but itâs like, how many people is putting this kind of movement into place?
00:38:16 Bryan Catanzaro: Yeah. Well, itâs, itâs, itâs hard to know exactly, and as I said, NVIDIA is a company of volunteers. But and also these days, things are changing, right? Like, so the Parakeet team, which is an excellent team, by the way they I would say a year ago wouldnât have really considered themselves so much part of the core Nemotron effort, but these days they absolutely are. for the obvious reason that, you know, LLMs these days need to be able to consume all sorts of data, right?
Including audio data. And so you know, as the pro-- as the characteristics, the capabilities of Nemotron models expand obviously, the number of people contributing is gonna expand. Iâd say right now thereâs about five hundred people that are working pretty much full-time on Nemotron technologies in different ways. This is everything from numerics quantization recipes to speech recognition or image understanding or, you know, pre-training, post-training, RL systems inference software. you know, thereâs, thereâs a, thereâs a whole bunch of different dimensions, right?
So Iâd say itâs about five hundred people. but also weâre having our Nemotron all-hands meeting this week, and so I took a look to see how many people were invited to that all-hands meeting, and it was about two thousand. so those are people around the company that are interested in working with Nemotron and either expanding its capabilities or helping its adoption. and so I think you know, the number is somewhere in between and itâs hopefully gonna keep growing as, as Nemotron matures.
00:40:07 Nathan Lambert: Yeah, I mean, thatâs one of the greatest attestations to what youâre saying is like, if the interest outside the company-- inside the company is four times as big as the people doing it, youâre gonna, youâre gonna keep scaling up, it seems. People are gonna-.. find ways to help. - One of the other things Iâm interested in, I donât know, like, on the point of five hundred, itâs like, it sounds like a lot of people, but with how many things you have going on, it seems also very few. âCause Iâm transitioning to thinking about the long-standing, like, open-source software that youâve had for NeMo, and I think Megatron, and itâs like theyâve been around for a long time. I think Megatron has gone through many eras. I have a note here.
Itâs like these softwares have been going around since, like, twenty nineteen in some form. And itâs, it-
00:40:51 Bryan Catanzaro: Publicly. We had our first public release in twenty nineteen, but we started earlier.
00:40:56 Nathan Lambert: And itâs something that Iâve found is that when I started doing lang- language models, so I was a late bloomer, and weâll transition to some career talk in a few minutes at Hugging Face. Like Megatron had, like, a bad rap of being very hard to use. But now, like three years later, I hear from anyone thatâs founding a new language modeling startup, theyâre like, âJust use Megatron.â like, do you pick up on things like this? Is it just, like, random-
00:41:22 Bryan Catanzaro: Well, we-
00:41:22 Nathan Lambert: .. but itâs like-
00:41:22 Bryan Catanzaro: We hard on it. You know, weâre trying really hard to make Megatron easier to use. Itâs difficult. Megatron is a complicated piece of technology, and, you know, when we originally started Megatron, the point was to show the community that you could make state-of-the-art large transformer language models with NVIDIA.
I donât know if you recall, but it-- there was some assertions by some other companies back in twenty seventeen when the transformer was invented, that they could only be made without NVIDIA. in fact, there were statements to that effect on bl-- on official blog posts, which I think got redacted later on. But it was important for NVIDIA to show up and say, âWe love language models. We love transformers. Letâs see what we could do, you know, if we partitioned the work properly on lots of GPUs with an amazing interconnect, what kinds of models could we train?â And so thatâs where the Megatron project started.
You know, I actually came up with the name Megatron. one of my proudest moments, I suppose. I was thinking about it, I was like: This is a really big transformer. Whatâs the biggest and baddest transformer? Oh, itâs Megatron.
So thatâs, you know, where the name came from. but youâll think about that had nothing to do with usability, right? Like, I wasnât, I wasnât thinking about, like, how do we make a platform thatâs really easy for other people to use? I was just trying to show the world that, like, NVIDIA systems could be awesome for transformers. You know, that was, that was my goal.
Over the years, you know, it has evolved. We have a lot more people trying to use Megatron. We got a lot of complaints about how hard it was to use, and then we did a lot of work to try to improve the software engineering around Megatron. You know, these days Megatron software engineering is actually shared between about four different teams at NVIDIA. and we have to coordinate that work very closely.
That has also not been easy. There has been times when you know, people wanted to fork Megatron, and then there were times when we, like, had to bring it back together, and itâs like: Look, I know forking things is always tempting, but look, better together. Itâs better for all of us to keep working together.. and so I feel like Megatron the-- and especially Megatron Core, which is like a subset of Megatron thatâs, like, especially protected, and we try to put more software engineering into that that has gotten dramatically better since we started paying more attention to it as a company. are we done yet? No, thereâs a lot, a lot, a lot more work.
00:43:52 Nathan Lambert: a ba-- a basic question: Is is Megatron or Megatron Core, like, this is what Nemotron is trained on? And also-- And itâs also something that many of the hottest, like, AI startups are training their models on. I would guess that thereâs nothing else that does that. So, like, could you summarize why itâs so hard?
00:44:11 Bryan Catanzaro: Well, you know, thereâs a, thereâs a lot of other great frameworks out there. Megatronâs not the only one. and you know, weâre happy about that. NVIDIA doesnât need to control the space. What we, what we do wanna do is make sure that weâre putting our products forward in the best light, you know, and itâs a challenging problem.
Weâve got so many things going on with precision and you know, the networking. Like, those questions, like, the software is so complicated. these days, you know, weâre pre-training our Nemotron-3 Super and Ultra models using FP4 which is a thing that, you know, hasnât been done publicly anyway and something that, you know, weâre pretty excited about because our GPUs have really awesome FP4 throughput. But obviously, the numerical challenges of, like, trying to train a state-of-the-art language model using four bits is non-trivial. So, like, you know, all of that work has to go into Megatron, into Transformer Engine which is a, another open-source project that Megatron relies on and, you know coordinating all of that making sure that, you know, we can actually deliver the benefits of NVIDIA systems to people that are trying to make state-of-the-art models, thatâs really important to us.
And, you know, of the five hundred or so people working on Megatron, like, a pretty good fraction.. or on Nemotron, a pretty good fraction of them are working on these kinds of systems issues, right? Because NVIDIA at its core, is a systems company. and Megatron, you know, Nemotronâs first job really is about systems, you know, and so we, we care, we care deeply about that.
00:45:51 Nathan Lambert: Yeah. I mean, from my perspective, I was at Hugging Face before AI2, and Hugging Face is, like, the best company at doing public work. But also, and switching to AI2 and focusing on, like, weâre focused on the output artifact the most. Seeing the different type-- Like, itâs such a different type of work, going from youâre trying to build a tool thatâs good for training models, to build a tool thatâs good for everybody else and whatever heck use case they are.
00:46:13 Bryan Catanzaro: Itâs different.
00:46:13 Nathan Lambert: So I think-
00:46:13 Bryan Catanzaro: Yeah. Different work.
00:46:14 Nathan Lambert: To do both is like.. Iâm, Iâm happy that AI2âs repos arenât that popular in terms-
00:46:21 Bryan Catanzaro: Oh,
00:46:21 Nathan Lambert: .. of open-source adoption because, like, we canât handle it. We just canât. Itâs, like, so hard because itâs people-- itâs, like, it ends up being researchers that are supporting it, and we donât have the ability to scale the organization structure. So I just think, like, thatâs a, thatâs a very fun turnaround for me to think of all these things happening at once.
00:46:39 Bryan Catanzaro: Yeah. Well, thanks for noticing weâre putting effort in. I would say Megatron is still not nearly as user-friendly as Hugging Face libraries. Like-.. Hugging Face libraries are legendary, and I admire the work theyâve done to make the community so productive. people, you know, are able to get so much research done thanks to the work that, you know, Hugging Face has put into to their library. So you know, my hatâs off to them as well.
00:47:06 Nathan Lambert: Yeah. One of my hot takes, you donât have to reply, is that Hugging Face and NVIDIA have been very good partners.
00:47:10 Bryan Catanzaro: Oh, absolutely.
00:47:10 Nathan Lambert: And itâs like bringing that Hugging Face culture to the NVIDIA stuff would be so good. Itâs just so hard, so I donât know how that would work, but-
00:47:17 Bryan Catanzaro: Weâre trying, you know, and you know, it is, it is challenging. NVIDIA is always a company that is gonna prioritize speed like hardware speed, above really anything else, âcause thatâs, like, who we are. I am always trying to make the case that developer speed is important, too, right? Itâs like thereâs different ways of thinking about speed. and it is definitely the case that a lot of NVIDIAâs software is so cumbersome to use that you know people canât get the actual hardware speed as fast as it should be because they just give up.
You know, they just donât, donât even figure out how to use that. So I think NVIDIAâs making strides there. I think the, the company is understanding more deeply how important developer experience is, and I hope we continue to push that, so that the benefits of all of the systems technology that NVIDIA works so hard on can be more widely used. but at the same time, you know, there is gonna be a tension between those things. Itâs, itâs not gonna go away, and you know, to a certain extent, I think thatâs just life on planet Earth.
00:48:26 Nathan Lambert: It is. I think youâre do- youâre doing a good job, and Iâm gonna kind of shift gears in this interview. So Iâve.. In becoming more back in language- in becoming a person that works in language models, Iâve seen your name more and more times.
I was like, âBryan Catanzaro, like, where have I seen this?â And then I went and did the research of the Berkeley PhD in, like.. It says April of 2021, you gave a Berkeley EECS Colloquium titled âApplications of Deep Learning and Graphics, Conversational AI, and Systems Design.â Iâm not even gonna posit that I actually went, but thatâs definitely where I remembered the name from in grad school. And we both have backgrounds that arenât traditionally in AI and end up working in language models. I just wanted to, like-- what have you learned from your path th- through NVIDIA into what, like, people should be thinking about with AI or open models today?
This could be career reflections, like technical reflections. I just think that thereâs-- there are actually a lot of people that come from all over the, like, STEM field to work in AI, so giving it-
00:49:29 Bryan Catanzaro: Sure
00:49:29 Nathan Lambert: .. space to think about is-
00:49:31 Bryan Catanzaro: .. useful, even if itâs just like, it was the big problem, and I wanted to go solve it. Well, I think, you know Iâve, Iâve had a lot of opportunity and a lot of luck in my career. I think in hindsight, it seems like an extraordinarily lucky thing that, you know, I did my first internship at NVIDIA in 2008, and I was, like, building machine learning models on the GPU, and I went to NVIDIA, and nobody else was really doing that. And I was like, âHey, like, we should have more people doing machine learning on the GPU.
I think this could be an opportunity.â And you know, it took a few years for me to make any headway. NVIDIA didnât really wanna listen to me. I was a brand-new PhD. I was in the research organization, which is very independent, but, you know, sometimes struggles to change the way that the, you know, the bigger company thinks about things.
And and yet, I just had this conviction, you know, I just was following my heart about what I think is gonna be important, what do I think could really change the world? And that has been, I think, the thread that has taken me through my whole career, is that Iâm constantly trying to refine my beliefs about what matters and then hold to them. And that.. I donât know how helpful it is to say that, but I feel like sometimes people you know, tend to follow the, whatever the thing is that people are talking about on Twitter.
And like Iâve- Iâve done a lot of unpopular things during my career because I believed in them, you know? I remember I published my first paper in 2008 on, at ICML, on training support vector machines on the GPU, and I actually had somebody at the conference, it was in Helsinki at dinner, you know, we were all telling each other what weâre doing, and, and I was like: Yeah, I wanna help people train bigger models on bigger data sets with GPUs. And, and I had you know, a couple of people just say, âWell, why are you here at ICML? That just doesnât really feel like a good thing for us.â And in 2008, ICML was momly- mainly about new mathematical frameworks for thinking about data, and you know, maybe if you trained a model at all, you would train one on your laptop.
You know, that was the state of machine learning in 2008. So for somebody to come in and say, âI think I want to focus on, like, parallel computing, new kinds of hardware for machine learning, programming frameworks for machine learning, so that, you know, we- more people can try inventing new models on complicated machines with a lot more compute throughput on bigger data sets,â that was like a, an unpopular thing. At least it felt very unpopular. I felt very marginalized at the time by the community.
But I believed in it, you know? I just felt like, look, technology.. Like I have this sense of, like, where do I think technology is going? I knew that traditional computing was running out of steam.
You know, I had, I had done a few internships at Intel, and I was trying to help Intel make processors that ran at, like, ten gigahertz back in 2001, and, you know, it was, like, clear that th- they were running into a wall. And I was thinking: Okay, so if the compute hardware is gonna have to be different, itâs gonna be more restricted. Itâs not gonna be able to be so general-purpose in order to get speed. What kinds of applications are gonna have, like, an infinite need for more computing?
And I thought, well, machine learning and AI, that could really change the world if it ever actually worked. But, you know, but, you know, back then it, back then, it kinda worked inside of Google. outside of Google, it kind of didnât work. and so I had kinda these signals, like it was possible, but it was hard. It was a little weird. It was a little niche.
I was a little bit caught in between different fields, like the systems people didnât think I was systems enough, and the machine learning people didnât think I was machine learning enough. But, but I believed in what I was doing, and I found a way to keep following that belief. And, you know, ultimately it was very rewarding when all of a sudden NVIDIA decided, âHey deep learning is changing the world. What do we know about deep learning?â And then it was like: Oh, well, Bryanâs been doing that for several years, and heâs written some libraries that we could turn into a product.
Letâs go do that. And, you know, so that all happened really quickly after many years of nothing happening, you know? And that was really obviously an amazing opportunity for me. you know, an- another thing that was important to me, I left NVIDIA in 2014 to go work at the Silicon Valley AI Lab at Baidu with a group of really talented people, including Andrew Ng and Dario Amodei and Awni Hannun and Adam Coates, and you know, this was a, a really once-in-a-lifetime opportunity, I think for me, to learn some things that would have been hard for me to learn on my own. you know, I felt at the time at NVIDIA that although I had this great opportunity to help NVIDIA become an AI company, and I was doing that, and I was succeeding at that back in 2013 2014, I also felt like I really wanted to learn from a broader community of people applying machine learning and AI to solve really important business problems. And so going to work at Baidu really gave me that chance. and I was there for a couple of years, learned a ton. very grateful to the team there especially to Andrew Ng, who, who encouraged me to, to join with him on that. and then, you know, I ran into limits of what I could do in California, working for a Chinese company.
I was thinking about, you know, what should I do next? And Jensen asked me to come back and build an applied research lab at NVIDIA in 2016. and -.. I wasnât sure, like, if that was a good idea. I thought NVIDIAâs already grown so much, you know.
The, the years from twenty fourteen to twenty sixteen, NVIDIA actually grew a lot. these days you look back at it, and youâre like: It was still really tiny. But, but back then, I was like: I donât know, maybe NVIDIAâs already tapped out. I donât know if you recall, in twenty sixteen, there was already, like, ten different companies making GPU competitors, right? The TPU had already been out for a while and you know, it, it wasnât clear that NVIDIA was gonna become as large as it, as it has.
But I believed in the opportunity. I believed in the people. you know, one of the things I loved about NVIDIA was that itâs a very stable organization. So Jensen, heâs been running it since he founded it in nineteen ninety-three. my boss, Jonah Alben, whoâs an absolutely extraordinary person has been here for you know quite a, quite a long time, almost since the very beginning of NVIDIA. And these people a lot of the leadership at NVIDIA they love the work.
Their heart is in the work. Jensen and Jonah and many other leaders at NVIDIA, they donât need to be doing this, right? They, they have earned the right to go sit on a beach and drink mai tais all day, but their heart is in the work, and they work incredibly hard. you know, the.. I feel like if there was an Olympics for email, you know Jensen would get the gold medal.
You know, like itâs, itâs unfathomable to me, like, how much information heâs able to process. and itâs a skill that heâs built up over a long time running this company, but itâs also a reflection of his commitment to the work. And I felt like working at a place where weâve got this very stable organization that loves the work, that really wants to change the world. You know, why does, why does Jensen get up in the morning? Well, itâs-- this is his chance to do something meaningful.
I thought, associating with these people, you know, I could do worse. I could-- I think I could learn from this as well. And so I came to NVIDIA, and back then it was really hard to explain to people why I was trying to build an AI lab inside of NVIDIA. At, at the time, NVIDIA wasnât doing very much AI, and so I had to kind of develop a vision for that and then explain it to people. thatâs ended up being a really good idea for me as well.
You know, the lab, I think, has really helped NVIDIA. you know, Megatron, I think, has really shown the industry, like, how valuable NVIDIA systems can be for language modeling, which is, which is awesome. DLSS, you know Iâm continuing to, to push DLSS forward. Very excited about making graphics, you know more efficient with AI. These days, you know, fifteen out of every sixteen pixels a gamer sees are rendered by AI models that, you know, my team developed, and that then makes the GPU ten times more power efficient.
This is a really exciting you know, thing for me to be involved with, something that Iâve, you know, dreamed about for years. So, so thatâs the kind of thing that continues to push me forward, is that I have strong beliefs about what I think is possible, where I think technologyâs going, and Iâm willing to do things that are we- weird and unpopular but, you know, basically following my convictions. Iâm very much always thinking about the people Iâm working with, the tribe. You know, I think tribes matter enormously. like you know if I..
So, so back when I was a grad student, I was working on programming models for machine learning. I joined the Python tribe. There are other people that were in the Scala tribe, and the people that did their work in the Scala tribe, trying to make programming models for machine learning in, like, two thousand and ten you know, that work, although a lot of it was technically excellent, didnât matter to the community as much as the people who were in the Python tribe. It ended up.. and, you know, it kind of sucks sometimes that the world is tribal like this, but itâs just the case.
You know, that like the people that you work with, the community that you work with has a big impact on the problems you think about and then the impact that your work has. So I think a lot about the people and the tribes that Iâm collaborating with or that Iâm part of. and you know, thatâs, thatâs kind of been the thread that has carried me through my career.
00:59:56 Nathan Lambert: Yeah. Than- thanks for sharing this full arc. I think youâve said things that I tell people but in different languages, and the first one, the early days, it seems like there can be space in between fields, where people-- two fields will have their way of describing things, but both of them are probably incomplete, and there can be space there, which is a lot of what I was doing transitioning from novel robots to model-based RL, where I, like, didnât sit and bear in the actual AI lab, but I started doing AI with my, like, total electrical engineering friends. And then the second thing is, like, Iâd wholeheartedly recommend this to people, is, like, choose your work based on the people and people that sincerely are in it for-.. the, what they want to do, and a lot of-
01:00:41 Bryan Catanzaro: And follow your beliefs. You know, think about it. What do you believe in? And itâs okay to change your mind, you know, but, like, figure out what is it that you believe in.
Ask yourself every day: Do I still believe in that? If I do, what next? You know. If I donât, well, what do I believe in?
You know, thatâs been really important to me. I think too many people end up kind of just following trends. Thatâs not usually helpful because the trends are too late. So if you wanna, if you wanna change the world, you need to be ahead of the trends, and you need to know, you know, it-- trends-- I donât think trends in computing are just fashion.
I think thereâs truth that drives those trends. Not always, but often. You know, itâs just-- this is, itâs thereâs kind of an inevitable force of gravity. It just can be really hard to par- parse out the noise and figure out what is the truth that is gonna push the industry forward, and how can you push that with it.
You know, if you can join with that, you can accomplish great things.
01:01:36 Nathan Lambert: Yeah, I agree. I think in building language models, itâs like you want to build a model that the community wants in six months. I think if youâre building a model to compete-.. with the models that are already out, youâre not gonna keep up. And I think that itâs like, what is the right thing is building open language models in six months, and like, where do you need to try to steer things is one of the hardest problems that I think about. So I donât-- if you want to close with any predictions where you see, like, open models, like, if weâre-- if youâre gonna be here at the end of twenty-six, if thereâs anything you think will be far more obvious than it is today, or any bets that you want to make, I think itâs kind of a good place to wrap.
01:02:18 Bryan Catanzaro: Well predictions are always hard, and I donât feel like Iâm very good at making predictions. But I am-- I feel like I am good at identifying what I believe in, and what I believe in right now is that compute remains one of the fundamental challenges behind AI. It has been that way for a very long time and I think it continues to be. I think as we find new ways to apply compute to AI, we discover new forms of scaling laws that help AI become more useful and therefore, it becomes more widespread.
So Iâm gonna keep thinking about compute. I continue to believe that the fastest-- that, you know, the way to think about AI is not just in terms of absolute intelligence, but rather intelligence per second. You know, thereâs some sort of normalization in there that relates to how fast a model can think, how fast a model can be trained or post-trained. You know, that models that kind of incorporate this compute acceleration characteristic, where theyâre thinking about intelligence per unit time, those are gonna end up winning because they end up getting trained on more data, they end up getting post-trained with more cycles, they end up with more iterations during thinking when theyâre deployed. and you know, of course, if they happen to fit the hardware really well whatever hardware that is then, you know, that can have a pretty non-trivial effect on the intelligence as well.
So thatâs something that I really believe in. I really believe in AI as an infrastructure. You know, thereâs, thereâs different ways of thinking about AI. I think some people believe AI is more like the singularity, like once AGI has been declared, then the whole world is different forever, and all humans have lost their jobs and, you know, thereâs a lot of like-- thereâs a lot of things about AI that people believe that I personally donât believe.
You know, I believe, first of all, that intelligence is very multifaceted that it is not easy to pin down, that as soon as we try to pin down intelligence, we find that thereâs very many more forms of intelligence that arenât covered by that. So, for example, a model that achieves gold medal status on the International Math Olympiad, thatâs an extraordinary achievement, but it doesnât make me have no job, right? Like, Iâm actually not solving math problems all day, even though, like, having the ability to solve math problems is clearly very useful. And you know, itâs also the case that intelligence is, you know, is kind of like a potential energy itâs not a kinetic energy, right?
In order to transform intelligence into kinetic energy, it needs to have a platform. It needs to be applied in the proper way. and you know, that is why I believe in open models and open- openly developed and deployed intelligence. I believe every company, every organization, has secrets that only they know. They have special data, they have special ways of thinking about their problems, their customers, their solutions, and theyâre gonna know how to apply AI better than anyone else.
And so AI as infrastructure that transforms companies, turbocharges them, allows them to take the things they know and multiply their impact, thatâs something that I believe in more than AI as an event, that one day, when it happens, makes everyone obsolete. I donât.. I just donât believe in that. you know, I often joke that, like if, for example, the CEO were to retire at some point, and we needed to find a replacement you know, handing out an IQ test or asking, you know, who has the highest SAT score that would not be a very good way of finding a replacement, you know? intelligence is just far too complex for that. And so you know, so this, these beliefs, you know, you can disagree with me about anything that I just said, and Iâm not offended by that.
I have a lot of friends that do. but you know, Iâm asking myself, well, if I believe that intelligence has these characteristics and that AI is gonna change the world by turbocharging institutions that exist a-and also creating new applications that we havenât even dreamed of yet rather than replacing all humans, then, you know, how do I go about building that, you know? And so thatâs, thatâs kind of the direction that Iâm on right now.
01:07:00 Nathan Lambert: Yeah, I love it. I agree, I agree that weâre entering an interesting area where the open models are taking so many different shapes and sizes and have so many different strengths and trade-offs, that there can start to be interesting interplay as an ecosystem, where thereâs just so many different things going on. And I think I like your idea of potential energy, and you have to build things that are kind of unclear of what-- Itâs like you have to build the energy in a way, and you donât really know what the goal is, but you have to do.. try to build these good models. So I appreciate it, and-
01:07:30 Bryan Catanzaro: Yeah, and then let people apply it. Let it-- let them make the kinetic energy happen.
01:07:35 Nathan Lambert: I agree. Thanks for coming on.
01:07:37 Bryan Catanzaro: Thanks so much for inviting me. Itâs been a great conversation.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Thereâs a pervasive, mutual challenge in the job market today for people working in (or wanting to work in) the cutting edge of AI. On the hiring side, it often feels impossible to close, or even get interest from, the candidates you want. On the individual side, it quite often feels like the opportunity cost of your current job is extremely high â even if on paper the actual work and life youâre living is extremely good â due to the crazy compensation figures.
For established tech workers, the hiring process in AI can feel like a bit of a constant fog. For junior employees, it can feel like a bit of a wall.
In my role as a bit of a hybrid research lead, individual contributor, and mentor, I spend a lot of time thinking about how to get the right people for me to work with and the right jobs for my mentees.
The advice here is shaped by the urgency of the current moment in LLMs. These are hiring practices optimized for a timeline of relevance that may need revisiting every 1-2 years as the core technology changes â which may not be best for long-term investment in people, the industry, or yourself. Iâve written separately about the costs of this pace, and donât intend to carry this on indefinitely.
The most defining feature of hiring in this era is the complexity and pace of progress in language models. This creates two categories. For one, senior employees are much more covetable because they have more context of how to work in and steer complex systems over time. It takes a lot of perspective to understand the right direction for a library when your team can make vastly more progress on incremental features given AI agents. Without vision, the repositories can get locked with too many small additions. With powerful AI tools I expect the impact of senior employees to grow faster than adding junior members to the team could.
This view on the importance of key senior talent has been a recent swing, given my experiences and expectations for current and future AI agents, respectively:
Every engineer needs to learn how to design systems. Every researcher needs to learn how to run a lab. Agents push the humans up the org chart.
On the other side, junior employees have to prove themselves in a different way. The number one defining trait I look for in a junior engineering employee is an almost fanatical obsession with making progress, both in personal understanding and in modeling performance. The only way to learn how the sausage gets made is to do it, and to catch up it takes a lot of hard work in a narrow area to cultivate ownership. With sufficient motivation, a junior employee can scale to impact quickly, but without it, itâs almost replaceable with coding agents (or will be soon). This is very hard work and hard to recruit for. The best advice I have on finding these people is âvibes,â so I am looking for advice on how to find them too!
For one, when I brought Florian Brand on to help follow open models for Interconnects, when I first chatted with him he literally said âsince ChatGPT came out Iâve been fully obsessed with LLMs.â You donât need to reinvent the wheel here â if itâs honest, people notice.
For junior researchers, thereâs much more grace, but thatâs due to them working in an education institution first and foremost, instead of the understatedly brutal tech economy. A defining feature that creates success here is an obsession with backing up claims. So a new idea improves models, why? So our evaluation scores are higher, what does this look like in our harness? Speed of iteration follows from executing on this practice. Too many early career researchers try to build breadth of impact (e.g. collecting contributions on many projects) before clearly demonstrating, to themselves and their advisors, depth. The best researchers then bring both clarity of results and velocity in trying new ideas.
Working in academia today is therefore likely to be a more nurturing environment for junior talent, but it comes with even greater opportunity costs financially. Iâm regularly asked if one should leave a Ph.D. to get an actual job, and my decision criteria is fairly simple. If youâre not looking to become a professor and have an offer to do modeling research at a frontier lab (Gemini, Anthropic, OpenAI is my list) then thereâs little reason to stick around and finish your Ph.D.
The little reason that keeps people often ends up being personal pride in doing something hard, which I respect. Itâs difficult to square these rather direct pieces of career advice with my other recommendations of choosing jobs based on the people, as youâll spend a ton of your life with them, more than the content of what youâll be doing. Choosing jobs based on people is one of the best ways to choose your job based on the so-called âvibes.â
Working in a frontier lab in product as an alternative to doing a Ph.D. is a path to get absorbed in the corporate machine and not stand out, reducing yourself to the standard tech career ladder. Part of what I feel like works so well for me, and other people at Ai2, is having the winning combination of responsibility, public visibility, and execution in your work. There is something special for career progression that comes from working publicly, especially when the industry is so closed, where people often overestimate your technical abilities and output. Maybe this is just the goodwill that comes from open-source contributions paying you back.
If you go to a closed lab, visibility is almost always not possible, so you rely on responsibility and execution. It doesnât matter if you execute if youâre doing great work on a product or model that no one ever touches. Being in the core group matters.
This then all comes back to finding the people hiring pipeline.
There are many imperfect signals out there, both positive and negative. For individuals building their portfolio, itâs imperative to avoid negative signals because the competition for hiring is so high. A small but clear negative signal is a junior researcher being a middle author on too many papers. Just say no, it helps you.
The positive signals are messier, but still doable. Itâs been said that you can tell someone is a genius by reading one Tweet from them, and I agree with this. The written word is still an incredibly effective and underutilized communication form. One excellent blog post can signify real, rare understanding. The opposite holds true for AI slop. One AI slop blog post will kill your application.
The other paths I often advise people who reach out asking how to establish a career in AI are open-source code contributions or open research groups (e.g. EluetherAI). Iâve seen many more success cases on the former, in open-source code. Still, itâs remarkably rare, because A) most people donât have the hardware to add meaningful code to these popular LLM repositories and B) most people donât stick with it long enough. Getting to the point of making meaningful contributions historically has been very hard.
Doing open-source AI contributions could be a bit easier in the age of coding agents, as a lot of the limiting factors today are just bandwidth in implementing long todo lists of features, but standing out amid the sea of AI slop PRs and Issues will be hard. Thatâll take class, creativity, humanity, and patience. So, to be able to run some tiny models on a $4000 DGX Spark is an investment, but itâs at least somewhat doable to iterate on meaningful code contributions to things like HuggingFaceâs ML libraries (Iâve been writing and sharing a lot about how Iâm using the DGX Spark to iterate on our codebases at Ai2).
Back to the arc of hiring, the above focused on traits, but the final piece of the puzzle is alignment. The first question to ask is âis this person good?â The second question is, âwill this person thrive here?â Every organization has different constraints, but especially in small teams, the second question defines your culture. In a startup, if you grow too fast you definitely lose control of your culture. This isnât to say that the company wonât have a strong or useful culture, itâs to say you canât steer it. The culture of an organization is the byproduct of how all the individuals interact. You do not want to roll the dice here.
Interconnects AI is a reader-supported publication. Consider becoming a subscriber.
Personally, Iâm working on building out a few more spots in a core post-training methods team at Ai2. Post-training recipes have gotten very complicated, and weâre working on making them easier to run while doing research on fundamentals such as post-training data mixing and scaling laws. To be a little vague, getting the post-training recipes done for both Olmo 3 and Olmo 2 was... very hard on the team. At the same time, post-training hasnât gotten much more open, so hiring through it and doing the hard work is the only way.
Ideally I would hire one engineer and one researcher, both fairly senior, meaning at least having a Ph.D. or a similar number of years working in technology. Junior engineers with some experience and the aforementioned obsession would definitely work.
This callout serves as a good lesson for hiring. It is intentional that people should self-filter for this, no one likes when you way overreach on selling yourself for a job. I also intentionally make people find my email for this as an exercise. The art of cold emailing and approaching people in the correct pipelines is essential to getting hired. Many people you look up to in AI read their emails, the reason you donât get a response is because you didnât format your email correctly. The best cold emails show the recipient that they learned from it or obviously benefitted from getting it. Platitudes and compliments are of course nice to receive, but the best cold emails inspire action.
Two of the most recent people I helped hire at Ai2 I learned of through these side-door job applications (i.e. not found through the pile of careers page applications). I learned of Finbarr through his blogs and online reputation. Tyler sent me an excellent cold email with high-quality blog posts relating to my obvious, current areas of interest and had meaningful open-source LLM contributions. Both have been excellent teammates (and friends), so Iâm always happy to say the system works, itâs just intimidating.
All together, Iâm very torn on the AI job market. Itâs obviously brutal for junior members of our industry, it obviously feels short sighted, it obviously comes with tons of opportunity costs, and so on. At the same time, itâs such a privilege to be able to contribute to such a meaningful, and exciting technology. My grounding for hiring is still going to be a reliance on my instincts and humanity, and not to get too tied down with all the noise. Like most things, it just takes time and effort.
Other posts in my âlife thoughtsâ series include the following. I send these to people when they ask me for career advice generally, as I donât have time to give great individual responses:
* Apr 05, 2023: Behind the curtain: what it feels like to work in AI right now
* Oct 11, 2023: The AI research job market s**t show (and my experience)
* Oct 30, 2024: Why I build open language models
* May 14, 2025: My path into AI
* Jun 06, 2025: How I Write
* Oct 25, 2025: Burning out
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe -
Arcee AI is a the startup Iâve found to be taking the most real approach to monetizing their open models. With a bunch of experience (and revenue) in the past in post-training open models for specific customer domains, they realized they needed to both prove themselves and fill a niche by pretraining larger, higher performance open models built in the U.S.A. Theyâre a group of people that are most eagerly answering my call to action for The ATOM Project, and Iâve quickly become friends with them.
Today, theyâre releasing their flagship model â Trinity Large â as the culmination of this pivot. In anticipation of this release, I sat down with their CEO Mark McQuade, CTO Lucas Atkins, and pretraining lead, Varun Singh, to have a wide ranging conversation on:
* The state (and future) of open vs. closed models,
* The business of selling open models for on-prem deployments,
* The story of Arcee AI & going âall-inâ on this training run,
* The ATOM project,
* Building frontier model training teams in 6 months,
* and other great topics. I really loved this one, and think you well too.
The blog post linked above and technical report have many great details on training the model that Iâm still digging into. One of the great things Arcee has been doing is releasing âtrue base models,â which donât contain any SFT data or learning rate annealing. The Trinity Large model, an MoE with 400B total and 13B active tokens trained to 17 trillion tokens is the first publicly shared training run at this scale on B300 Nvidia Blackwell machines.
As a preview, they shared the scores for the underway reasoning model relative to the whoâs-who of todayâs open models. Itâs a big step for open models built in the U.S. to scale up like this.
I wonât spoil all the details, so you still listen to the podcast, but their section of the blogpost on cost sets the tone well for the podcast, which is a very frank discussion on how and why to build open models:
When we started this run, we had never pretrained anything remotely like this before.
There was no guarantee this would work. Not the modeling, not the data, not the training itself, not the operational part where you wake up, and a job that costs real money is in a bad state, and you have to decide whether to restart or try to rescue it.
All inâcompute, salaries, data, storage, opsâwe pulled off this entire effort for $20 million. 4 Models got us here in 6 months.
That number is big for us. Itâs also small compared to what frontier labs spend just to keep the lights on. We donât have infinite retries.
Once I post this, Iâm going to dive right into trying the model, and Iâm curious what you find too.
Listen on Apple Podcasts, Spotify, YouTube, and where ever you get your podcasts. For other Interconnects interviews, go here.
Guests
Lucas Atkins âX,LinkedIn â CTO; leads pretraining/architecture, wrote the Trinity Manifesto.
Mark McQuade â X, LinkedIn â Founder/CEO; previously at Hugging Face (monetization), Roboflow. Focused on shipping enterprise-grade open-weight models + tooling.
Varun Singh â LinkedIn â pretraining lead.
Most of this interview is conducted with Lucas, but Mark and Varun make great additions at the right times.
Links
Core:
* Trinity Large (400B total, 13B active) collection, blog post. Instruct model today, reasoning models soon.
* Trinity Mini, 26B total 3B active (base, including releasing pre-anneal checkpoint)
* Trinity Nano Preview, 6B total 1B active (base)
* Open Source Catalog: https://www.arcee.ai/open-source-catalog
* API Docs and Playground (demo)
* Socials: GitHub, Hugging Face, X, LinkedIn, YouTube
Trinity Models:
* Trinity models page: https://www.arcee.ai/trinity
* The Trinity Manifesto (I recommend you read it): https://www.arcee.ai/blog/the-trinity-manifesto
* Trinity HF collection â (Trinity Mini & Trinity Nano Preview)
Older models:
* AFM-4.5B (and base model) â their first open, pretrained in-house model (blog post).
* Five open-weights models (blog): three production models previously exclusive to their SaaS platform plus two research models, released as they shifted focus to AFM â Arcee-SuperNova-v1, Virtuoso-Large, Caller, GLM-4-32B-Base-32K, Homunculus
Open source tools:
* MergeKit â model merging toolkit (LGPL license return)
* DistillKit â knowledge distillation library
* EvolKit â synthetic data generation via evolutionary methods
Related:
* Datology case study w/ Arcee
Chapters
* 00:00:00 Intro: Arcee AI, Trinity Models & Trinity Large
* 00:08:26 Transitioning a Company to Pre-training
* 00:13:00 Technical Decisions: Muon and MoE
* 00:18:41 Scaling and MoE Training Pain
* 00:23:14 Post-training and RL Strategies
* 00:28:09 Team Structure and Data Scaling
* 00:31:31 The Trinity Manifesto: US Open Weights
* 00:42:31 Specialized Models and Distillation
* 00:47:12 Infrastructure and Hosting 400B
* 00:50:53 Open Source as a Business Moat
* 00:56:31 Predictions: Best Model in 2026
* 01:02:29 Lightning Round & Conclusions
Transcript
Transcript generated with ElevenLabs Scribe v2 and cleaned with Claude Code with Opus 4.5.
00:00:06 Nathan Lambert: Iâm here with the Arcee AI team. I personally have become a bit of a fan of Arcee, âcause I think what theyâre doing in trying to build a company around building open models is a valiant and very reasonable way to do this, âcause nobody really has a good business plan for open models, and you just gotta try to figure it out, and you gotta build better models over time. And like open-source software, building in public, I think, is the best way to do this. So this kind of gives you the wheels to get the, um... You get to hit the ground running on whatever youâre doing. And this week, theyâre launching their biggest model to date, which Iâm very excited to see more kind of large-scale MoE open models. I think weâve seen, I donât know, at least ten of these from different providers from China last year, and itâs obviously a thing thatâs gonna be international, and a lot of people building models, and the US kind of, for whatever reason, has fewer people building, um, open models here. And I think that wherever people are building models, they can stand on the quality of the work. But whatever. Iâll stop rambling. Iâve got Lucas, Mark, um, Varun on the, on the phone here. Iâve known some of them, and I consider us friends. Weâre gonna kind of talk through this model, talk through building open models in the US, so thanks for hopping on the pod.
00:01:16 Mark McQuade: Thanks for having us.
00:01:18 Lucas Atkins: Yeah, yeah. Thanks for having us. Excited.
00:01:20 Varun Singh: Nice to be here.
00:01:20 Nathan Lambert: What- what should people know about this Trinity Large? Whatâs the actual name of this model? Like, how stoked are you?
00:01:29 Lucas Atkins: So to- yeah.
00:01:29 Nathan Lambert: Like, are you, like, finally made it?
00:01:32 Lucas Atkins: Uh, you know, weâre recording this a little bit before release, so itâs still like, you know, getting everything buttoned up, and inference going at that size is always a challenge, but weâre-- This has been, like, a six-month sprint since we released our first dense model, which is 4.5B, uh, in, in July of last year, 2025. So, um, itâs always been in service of releasing large. I- itâs a 400B, um, thirteen billion active sparse MoE, and, uh, yeah, weâre, weâre super excited. This has just been the entire thing the companyâs focused on the last six months, so really nice to have kind of the fruits of that, uh, start to, start to be used by the people that youâre building it for.
00:02:16 Nathan Lambert: Yeah, I would say, like, the realistic question: do you think this is landing in the ballpark of the models in the last six months? Like, that has to be what you shop for, is thereâs a high bar- ... of open models out there and, like, on what youâre targeting. Do you feel like these hit these, and somebody thatâs familiar, or like MiniMax is, like, two thirty total, something less. I, I donât know what it is. Itâs like ten to twenty B active, probably. Um, you have DeepSeeks in the six hundred range, and then you have Kimi at the one trillion range. So this is still, like, actually on the smaller side of some of the big MoEs- ... that people know, which is, like, freaking crazy, especially you said 13B active. Itâs, like- ... very high on the sparsity side. So I donât actually know how you think about comparing it among those. I was realizing that MiniMax is smaller, doing some data analysis. So I think that itâs like, actually, the comparison might be a little bit too forced, where you just have to make something that is good and figure out if people use it.
00:03:06 Lucas Atkins: Yeah, I mean, if, if from raw compute, weâre, weâre roughly in the middle of MiniMax and then GLM 4.5, as far as, like, size. Right, GLMâs, like, three eighty, I believe, and, and thirty-four active. Um, so it-- you know, we go a little bit higher on the total, but we, we cut the, uh, the active in half. Um, it was definitely tricky when we decided we wanted to do this. Again, it was July when... It, it was July when we released, uh, the dense model, and then we immediately knew we wanted to kind of go, go for a really big one, and the, the tricky thing with that is knowing that itâs gonna take six months. You, you canât really be tr-- you canât be building the model to be competitive when you started designing it, because, you know, that, obviously, a lot happens in this industry in six months. So, um, when we threw out pre-training and, and a lot of our targets were the GLM 4.5 base model, um, because 4.6 and 4.7 have been, you know, post-training on top of that. Um, and, like, in performance-wise, itâs well within where we want it to be. Um, itâs gonna be... Technically, weâre calling it Trinity Large Preview because we just have a whole month of extra RL that we want to do. Um- But-
00:04:29 Nathan Lambert: Iâve been, Iâve been there.
00:04:31 Lucas Atkins: Yeah, yeah. But i- you know, weâre, weâre in the, um, you know, mid-eighties on AIME 2025, uh, GPQA Diamonds, uh, seventy-five, um, at least with the checkpoint weâre working with right now. Weâre still doing more RL on it, but, um, you know, MMLU Pro, uh, eighty-two. So weâre, weâre, weâre happy. Weâre really-- Like, for it being our first big run, like, just getting it trained was, was an extreme accomplishment, but then for it to actually be, like, a, a genuinely useful model is a, a cherry on top.
00:05:03 Nathan Lambert: Yeah, letâs go big picture. Uh, like, letâs recap. We have all of the... We have this full trinity of models. I think that thereâs a fun note. Uh, did I put it in this doc? Yeah, on Nano Preview, which was the smallest- ... youâre, like, charming and unstable. The model cardâs really funny. Um, ChatGPT, doing deep research on this, I was like, ChatGPT Pro just tagged next to it, âcharming and unstable.â And I was like: Is this a hallucination? And then in the model card, you have, like: âThis is a chat-tuned model with a delightful personality and charm we think users will love. Uh, we think- ... itâs pushing the boundaries, eight hundred million, um, active parameter, and as such, may be unstable in certain use cases.â This is at the smallest scale- ... which is like, I appreciate saying it as it is, and thatâll come up multiple times in the conversation. And then you have Mini, which is like, um, I think it was, like, 1B active, 6B total type thing. In my-- I, I donât have it, the numbers right in front of me. I have it somewhere else. Um-
00:05:52 Lucas Atkins: Yeah, Nano was, Nano was the 6B, uh, 1 active.
00:05:55 Nathan Lambert: Oh, yeah, yeah.
00:05:55 Lucas Atkins: And then, and the Mini was twenty-six, 3B active.
00:05:58 Nathan Lambert: Yeah. So, like-
00:06:00 Lucas Atkins: Um, yeah.
00:06:00 Nathan Lambert: -are these based on more of, like, you need to build out your training chops, or are you trying to fill needs that youâve-... heard from community, and like, I think for context, previously, your first open model was a base and post-trained model, which was Arcee 4.5B, which was a dense model- -which people like. And prior to that, you had, like, a long list of, like, post-training fine tunes that you had released. So before that, it was like a post-training shop, and I think that kind of history is i- important to fill in, âcause I think most people-- a lot of people are gonna meet you for the first time listening to this.
00:06:34 Lucas Atkins: Yeah, it, it, um, we chose those sizes for Mini and Nano, uh, specifically Mini, um, the 26B, 3B Active, because we wanted to de-risk, uh, large. Like, th- this has all been in service of getting to a model of, of, you know, the 400B class. So, um, we, you know, learned from doing the original 4.5B, that you might have everything on paper that you need to train a model, but i- inevitably, thereâs tremendous, you know, difficulties that come up, and, um, it, itâs-- we, we definitely knew we wanted to make sure that we, you know, solved some of... E- especially when it came to just doing an MoE model performance, uh, you know, like a, like an efficient, fast train of an MoE. So, um, we thought that that was a good ground where we could, you know, it wasnât crazy expensive, uh, but gave us a lot of data, uh, going into large. And then Nano just came about because we had some extra compute time, and we really want to do more research on, like, smaller models that are very deep. Um, and we hadnât really seen that in an MoE before, so that one was very much we started training it, and then it, you know, early benchmarks were good, so we said, âWell, weâll just do the whole dataset.â Um, and, uh, but most of the love for those releases went into, to Mini. So I, I definitely think that long term, uh, from an ROI perspective, the smaller models are going to be where we shine, just because thereâs a tremendous amount of, of cost savings a company can get from, from optimizing on a, on a smaller model. Um, but, but we, uh, w- weâre definitely gonna be trying to push the, the large frontier, too.
00:08:26 Nathan Lambert: Yeah. Um, Iâd like to kind of double-click on training before going back to the small model thatâs useful for companies, âcause weâre gonna have-- weâre gonna end up talking for, like, twenty minutes plus about open ecosystem. So I kind of am curious, like, philosophically, how your company feels about, like, sharing scientific details. So if I ask you, like, what are the things youâre technically most excited about in the model, or, like, what are the pain points? Like, uh, like, are you willing to talk about these things? Like, I- Do you feel like itâs kind of orthogonal to the company? Like, I feel like a lot of it is just, like, things that happen. I think your framing of all of this is in service of getting the big model going. And particularly, of, like, you have to be thinking about your model as landing in six months, is probably... Like, for people not training models, itâs hard to think about, âcause even I- ... like, Iâm thinking about trying to refresh our post-training stack for OLMo 3, and Iâm like, the thinking model, the, um, we are pretty SFT heavy right now, and it makes it not very dynamic in terms of the thinking time. But itâs just like, I canât see people deploying this model, or probably will have a hard time fine-tuning it. And itâs like to think about where tool use models are going in six months, like, seems pretty hard. Um, itâs a very hard task to do, so it takes a lot of gumption to actually set out and do it. So I, I would just appreciate the framing, kind of self-reflecting on what I go through. So if you have anything that you think was, like, particularly hard to actually land the six-month outlook, because you use Muon as an optimizer, or is it Muon? And some of these things. I think the data, itâs well known that Datology is cranking a lot of this, and you probably provide-- I think of it as like youâre kind of driving and working with these partners, and Iâm sure you provide a lot of feedback on whatâs working and whatâs not. So- ... anything youâre willing to share, I think itâs useful.
00:10:08 Lucas Atkins: Uh, I, I think, um, I mean, on the data side, like Datology, I-- at least for these models, that, that partnership has very much been almost an extension of our own research team. Like, weâve worked very closely with them, and, um, obviously, our modelâs doing well, you know, i- is, is, is good for them. So, um, but it, it-- there was definitely, you know, and you know this better than most, like, small-scale ablations, when you throw them at scale, sometimes, you know, uh, the-- i- it doesnât always turn out how you want. So there was quite a lot of iterating there to at least get the dataset we used for Large. Um, I, I would say that as far as looking out six months and then figuring out how we wanted to... Obviously, the big one was compute. We donât, um, you know, we, we never raised as, like, a foundation model company, so weâve ne- we havenât signed massive commits for, you know, thousands of GPUs before. Um, we didnât have a, a, a massive cluster that was always active, uh, for a lot of our post-training. So if they came before, um, you know, we had sixty-four, uh, H100s, that was pretty sufficient for that kind of work, but obviously, this necessitated quite a bit more. Um, but the first thing was-
00:11:29 Nathan Lambert: Thatâs still less than people would guess. Like, youâre releasing models- ... that werenât like, your models werenât catching national news, but people in the community knew about them. And, like, uh, i- I think of, like, Moondream when I think about that. Like, vik has- ... such little compute, and he puts it to so use. Like, you, like, see how successful he is? And he tells you that he has, I donât know, thirty... Like, l- it might be, like, sixty-four GPUs. Like, uh- ... thereâs, uh, uh, thatâs a whole separate conversation on building- ... actual good ML output on little compute. I, I should ta- I should chat with vik about this, but aside
00:12:03 Lucas Atkins: No, itâs, it is-- I think it was... Yeah, it, it, it was very much a gift going into the pre-training side because-... we were kind of already thinking, All right, how do we do the mu- you know, the most with the, the least amount of compute? But, um, you know, we-- it took us quite a while to get the cluster that we have been training large on, which is twenty-two thousand forty-eight B300s. Um, and once we figured out when we were going to get that, get access to that cluster, everything else kind of became clear as far as, like, timelines for Mini and Nano and, and when we wanted to do that. Uh, obviously, you know, five hundred and twelve H100s was easier to come across, um, for Mini and Nano. So once we figured that out, um, it really became, uh, this game of, okay, how can we find, like, the best research on the topic of, of pre-training, and what is kind of... What are the, the, the papers and publications that are coming out, um, that have enough potential and enough precedence, either because, uh, another lab used them, it comes from a reputable team, uh, the ablations and the, the evaluation setup, like in the paper, was sufficient enough to give us confidence. Uh, and then we basically spent, I donât know, it was probably about two months just figuring out what we wanted our architecture to be for the MoE, then figuring out, okay, now that thatâs what we want to do, how do we implement all of that in the actual training pipeline? Uh, how can we-- you know, at that time, there had been many people whoâd done Muon, but, um, for post-training, and, and then other-- some Chinese labs had used it, but there wasnât, like, a widely available distributed Muon, um, to do it that scale.
00:13:54 Nathan Lambert: What do you think that, like, looks like in decision-making? âCause that seems like a risky decision, if you ask me. I think for one, the ti-
00:14:00 Lucas Atkins: Muon?
00:14:00 Nathan Lambert: ... the timing, the, the, like, timing sharing that youâre saying is good. Like, you said this for two months, and then, like... But, like, even Muon is like, thatâs a bet that would even take-- like, somewhere like AI2, that would take some serious evidence to go with it. We would want to ablate it. So like- ... on a single track, itâs like y- you had probably had a process for becoming fairly confident in it then.
00:14:24 Lucas Atkins: It- yes, but it, it was also, like, Kimi had, had just come out, and we knew that that one used Muon, and so we knew that it, at least, if implemented correctly, could deliver a good model. There werenât outstanding ablations done around like... You know, there wasnât a Kimi scale model done with Adam, and then compared to Muon and see the difference. But, um, that at least gave us enough confidence that if-
00:14:50 Nathan Lambert: What does Muon give you? Does it give you, like, memory saving, uh, in-
00:14:55 Lucas Atkins: No, itâs actually a little bit more memory. Itâs, itâs, itâs mostly-
00:14:58 Varun Singh: Itâs, uh-
00:14:58 Lucas Atkins: ... like the loss converges a bit quicker.
00:15:00 Varun Singh: Itâs, itâs less memory, actually. Itâs, uh, uh, only one momentum buffer instead of Adamâs two, uh, beta buffers, and then itâs also better convergence.
00:15:10 Nathan Lambert: Okay. So itâs, like, mostly designed around convergence, and then I know the math is different, which is where this momentum term changes.
00:15:15 Lucas Atkins: Well, it, it kind of came out... I mean, it had its, its, its big, you know, uh, explosion of popularity in the kind of nanoGPT speedrunning community. So it was kind of all built around converging to a certain, you know, validation loss faster, and, uh, that, that, that was, um... As for why we chose it as opposed to Adam, weâd used Adam for 4.5b, uh, but we also knew that if we wanted to move this fast, that we were going to have to make some pretty big bets, educated. Um, but, but still, we would have to make some, some, some risky decisions, um, beyond just, you know, training in general. So, um, there were a few that Muon we went with, uh, I think was, was one of our bigger bets. Uh, we ended up not doing, like, multi-token prediction or, or, or FP8 because we were throwing so many new things into the run at once, um, that-
00:16:12 Nathan Lambert: Do these apply for-
00:16:12 Lucas Atkins: ... if something were to go wrong-
00:16:13 Nathan Lambert: um, Mini and Nano? Are those also Muon, or are those- ... Adam as well? Okay, so then you- ... you get some de-risk from that. Do you know off the top of your head how many days it take to train each of those? Like, a, a good-
00:16:25 Lucas Atkins: Uh-
00:16:25 Nathan Lambert: ... ballpark for people, before-
00:16:27 Lucas Atkins: Yeah, so-
00:16:28 Nathan Lambert: going into the bigger run.
00:16:29 Lucas Atkins: So, so Mini, uh, so Nano on it was five hundred and twelve H200s, uh, took a little over thirty days. Um, and then Mini was about forty-five days.
00:16:45 Nathan Lambert: Okay. I think another thing- ... off the top of my head is I know that, like, a OLMo 1B dense would take us, like, eleven days on a hundred and twenty-eight H100s for a dense model. So, like, sixteen. So, like, the numbers- ... just go up from there. âCause then itâs like the question is like, Iâm guessing i- if those are forty-five days, and then you have-- you up the number of GPUs, itâs gonna be like a similar amount of time, or forty days for the big model, but much more stressful.
00:17:16 Lucas Atkins: Yeah, the big model was... But again, that was- we knew that we, we wanted- we felt confident that we could deliver a competitive and exciting model in January 2026. Like, we knew that it would-- we could... Who knows kind of where the research and what, what class and, and, and, and skill and performance of model is gonna come out in the next three months? Um, so we also knew that we really wanted to land sometime in January, and thatâs also why we also took- we went with B300s, even though definitely the largest public train of that size on B300s and, and the, um, you know, a lot of the software was not-- did not have, like, out-of-the-box B300 support. It was the only way we were gonna be able to train a model of this size in-
00:18:06 Nathan Lambert: Did you have to do this? Did you have to implement the... like, help solve version issues or other issues on B300s? âCause Iâve heard that-
00:18:13 Lucas Atkins: W-
00:18:14 Nathan Lambert: ... the rollout has been rough.
00:18:16 Lucas Atkins: We had to add-... a, a bit. There, there were a couple days where the, the data center had to take it offline to implement some bug fixes. It was, it was definitely, like, a very cool experience being on the bleeding edge, but, um, also, like, a little frightening âcause you just know, like, âOh, weâre not getting the most out of these that we possibly could.â So, um, a little bit of both.
00:18:40 Nathan Lambert: Uh, was your final training run stable, or did you have to do interventions through it?
00:18:46 Lucas Atkins: Uh, it was very stable, actually. Uh, it took-- the beginning of it was not. The, the, the first ten days were absolute, um... It, it would start very well and, and looked, you know, uh, the dynamics and the logs, and the graphs looked very similar to Mini and Nano, and then after, uh, around a trillion tokens, it- the- we- you know, youâd get collapsing, experts would start to go crazy. Uh, part of this is just, again, we are very sparse compared to what you, you, you have. So, um, you know, four hundred billion total, um, thirteen billion active, two hundred and fifty six experts. Like, it was, it was-
00:19:26 Nathan Lambert: Did you do a, uh, expert routing loss or some sort of balancing loss?
00:19:30 Lucas Atkins: Yeah. Yeah, yeah. Yeah.
00:19:32 Varun Singh: We did, um, we used DeepSeekâs, uh... We, we modified DeepSeekâs Auxiliary-loss-free, um, uh, loss balancing with our own, like, uh, with some tweaks, and then we also added a sequence loss like they, uh, did as well.
00:19:47 Nathan Lambert: Uh, was there Auxiliary-loss-free one from DeepSeek V3, or was that a later model?
00:19:51 Varun Singh: That was V3.
00:19:52 Lucas Atkins: It was V3.
00:19:52 Varun Singh: They did a separate paper on it as well. Yeah.
00:19:55 Nathan Lambert: Yeah. Yeah, that makes sense. I think a lot of people have derived from there. Um, have you- ... had issues on post-training as well? So I have a theory that the new algorithms weâre getting from the Chinese labs, like GSPO and SysPO, are primarily for problems that you solve when you have big MoEs and you have expert problems when trying to do the RL. And thatâs the whole reason that, like, I think our very serious AI two RL setup, like, weâre doing it on dense models, and weâre just like, âItâs fine. We donât have this big clipping problem, and as much like we donât have as much of a need to get the batch size as big to ac- activate all the experts.â So youâre saying you have so many experts and so much sparsity, that potentially sounds like youâre making RL harder.
00:20:36 Lucas Atkins: Um, yes. I will also... I will say that from just, like, a purely post-training side, we added as much as we po- we used- we... So our code base started from TorchTitan. Weâve had to make a ton of modifications to it to get it where we need it to be, but that was an excellent base. And from one of the bigger learnings from Mini and Nano was treating, uh, at least the SFT side of it, as a s- as a separate phase. Um, âcause with, with Mini and Nano, we finished the pre-training, we did context extension, then we took those and then ran those on, like, the sixty-four H100s we usually would do post-training on. Um, that presented a lot of challenges, uh, with the MoEs. They, they really... And thatâs kind of been a thing in the open space, is post-training MoEs, like, really, um, can be frustrating, even for SFT. So for Large, we added, uh, like, fine-tuning directly to TorchTitan, um, and did it all on the same cluster. So, um, from a performance standpoint, like, SFT was very, um... actually ended up being totally different.
00:21:42 Nathan Lambert: What is the actual difference between the q- the, the implementations then? Is it just kinda like you end up with different batch sizes and parallelism and stuff? Like why-
00:21:50 Lucas Atkins: Uh, I mean, we ended up, we... Yeah, we ended up needing to get it to do really, like, to get context parallelism really well, really good, âcause weâre obviously going at a higher sequence length, and then, um, just adding the proper loss masking. Um, it, it, it, it ended up being a relatively easy implementation, especially âcause we did all the pre-processing, uh, outside of TorchTitan.
00:22:13 Nathan Lambert: Interesting.
00:22:14 Lucas Atkins: Uh, and then on the RL side, yes, I would say itâs not, um, it didnât present itself as, as, as significantly harder than, than, um, Mini and Nano. However, that many GPUs does, so we didnât end up using, uh, two thousand of the B300s for that. That ended up being, uh, a thousand. So two, we just split the nodes in half.
00:22:39 Nathan Lambert: Yeah. That makes sense.
00:22:40 Varun Singh: On the dense model side of things, uh, you mentioned that you didnât need to use all the tricks and stuff. I, I think it is, uh... I think the, the, it- MoEs are just, in general, harder to RL, but I think itâs also, like, uh, b- because of, like, the KL mismatch between trainer and inference engine, right? Um, where you have, like, uh, sometimes the inference engine can pick different experts compared to, like, the trainer, uh, when you, like, do a forward pass on the same tokens. So I think there is definitely some, like, inherent instability with, with RL on MoEs.
00:23:13 Nathan Lambert: Yeah, that makes sense. Are, are... Okay, um, another question of, like, how much do you want to say? How do you feel about the state of public post-training recipes? Like, do you... Like, I, I feel like thereâs so little out there, and thereâs an opportunity to be seen as technical leaders by sharing just, like, more of what youâre doing. âCause I feel like weâve seen for years how complicated things can be, but also at, kind of at the same time... Like, we see this from the likes of Llama, has these really complicated recipes. But at the same time, I feel like just executing on a simpler recipe can get pretty close. But itâs just, like, very uns- I feel, uh, currently unsatisfied with how much I know about what are the actual core trade-offs of doing post-training well. And I think you could do a lot with SFT, but thereâs definitely, in this RL regime, more trepidation of kind of narrowing your model to either downstream use or, like, being able to do this multi-week RL run where you get the most performance.
00:24:06 Lucas Atkins: Yeah, I mean, I, I, from-- since RL has become such a pivotal part of the process beyond what, you know, DPO and, and, uh, and kind of your, your typical RLHF was in the past, like, we used to get quite, uh-... sophisticated with, with how we would do SFT and, and even our, our RL. We, we obviously, we make MergeKit, so we, we utilized merging, and we used to do a lot of distillation, um, to eke out as much performance as we could. Now that RL is such a massive part of the entire post-training stack, I, I have almost reverted us to just really solid but simple SFT. Um, like in, in large, I mean, weâve-- our post-training data set for, uh, Trinity Large is, uh, two hundred and thirty billion tokens. Like, like, it just like a really, really, really large-
00:25:09 Nathan Lambert: Thatâs ten X what we did. At least in SFT.
00:25:10 Lucas Atkins: And even that-- and even, even your tenant, like that was bef- before this kind of w- going at this scale and even kinda thinking and, and reasoning models. Like our largest SFT before that was five billion to-- weâd do, like, three epochs, but it was like five billion, you know, tokens, so- Um-
00:25:28 Nathan Lambert: Our non-reasoning model is, like, te- another ten X. So, like, our most latest instruct model is, like, two billion.
00:25:34 Lucas Atkins: Yeah, which is, uh, already a lot, you know. So, um, I, Iâve definitely... We-- you know, simplicityâs key because it also makes debugging anything easier, and then, um, devoting a lot of that sophistication to the RL. Our RL part is, like, really important. I do think that, I mean, the next, uh, phase of reinforcement learning for models of this scale is, is just scale. Is, is... Okay, we went from, you know, twenty billion SFT to two hundred and thirty, now weâre going from, you know, ten environments to a hundred. I think that that really is where youâre gonna get the biggest benefit. I also think thatâs why, you know, MiniMax and, and, and other players like GLM are so performant and just, like, have that extra bit of, of usefulness that goes beyond just what you see in the benchmarks, is theyâve, theyâve really embraced, like, long-form, uh, RL. And, and so, um, yeah, I mean, to be quite frank, our, our RL pipelineâs rather... immature might be the wrong word. Like, itâs, itâs, uh, thereâs definitely a lot more work we could do and a lot more work we need to do, but, um-
00:26:43 Nathan Lambert: Have you started the tool use side of RL?
00:26:46 Lucas Atkins: That-
00:26:46 Nathan Lambert: Or are you mostly... Well, um, beyond like, if youâre training on code, just verifying the code answer, I donât count yet as tool use. I would say, like, search and code integrated reasoning is what I think is gonna be like minimum table stakes, but do it- to do it well is really hard. Like, we have to, like- ... like, you, you really, like, uh... Thatâs what I want to do. I want all of our models to have that this year. Search is prob- you have to have, like, a partner to do search or just, like, illegally scrape Google if youâre gonna- ... youâre gonna serve this model onto a customer, and itâs gonna- ... what? Go, go to Google, like, what?
00:27:16 Lucas Atkins: Yeah. Yeah, no, I mean, I, I... Beyond, like, like, really kind of like long-form, like deep research or, um, you know, even like GPT-OSS style or, or G- GPT 5 style, where, you know, itâs doing a hundred tool calls before it gives you a response. Not there yet, um, but that is kind of... Once we get past the, the final kind of RL of Trinity Large, and, and we kinda look at where we go next, like, that is the next major hurdle, um, for sure, and itâs intimidating.
00:27:56 Nathan Lambert: How big is your, your team of- of... Like, how many people are spending the majority of their time on the model? And then I think we c- start to wrap up technical talk and zoom out a bit to ecosystem and company strategy.
00:28:09 Lucas Atkins: Uh, thereâs thirteen at Arcee- ... that are just, like, every, every single day is working on it. Yeah.
00:28:16 Nathan Lambert: And I guess thatâs a good number because these people are talking about data, but thereâs also, like, the whole data thing thatâs coming somewhere else. But also somebody else that wanted to pre-train a model, like they could just download the best fully open data set. And I donât think itâs gonna be quite as good, particularly in the fact that, um, like, if you look at OLMoâs models, we donât have a lot of tokens, so we need to, like, acquire- ... more tokens in the open still. But to, like, get a number of thirteen, where some are spending a bit of time on data, but thereâs the whole data abstraction, is actually kind of nice for somebody thatâs like... To do a serious modeling effort, you need to have this many people, I think.
00:28:50 Lucas Atkins: It, it was-
00:28:51 Nathan Lambert: Itâs reasonable to me.
00:28:52 Lucas Atkins: It was, it was a good number. I mean, I would say that, um, it, it was helpful to be able to, you know... This was like, how do we alleviate as many concerns as possible? Or how do we check off as many boxes, right? And itâs like, if weâre trying to do this in the shortest possible amount of time, like, we need to focus on what weâre good at, which is we- pretty good at post-training, and how do we get to the point where weâre able to do that? Well, we have to have a pretty strong base model. How do we get a strong base model? Weâll-- we have to, you know, figure out how to do it, perform, you know, efficiently across many, many GPUs, and then dataâs, you know, extremely important, so getting a partner that could, you know, help us with that, and we could offload some of that. It, it- there ended up being, obviously, as you, you know, alluded to earlier, like, a lot of, uh, working with Datology and, and, and others to make sure that the data accomplished what we needed it to. Um, I think that that is gonna be an interesting... You know, as we, as we- now that we have Large and weâre looking at, you know, kind of going further, itâs like, okay, you know, the, the pre-training data really has to be in service of what you wanna do in the post-training, uh, work.
00:30:10 Nathan Lambert: How did you identify this?
00:30:11 Lucas Atkins: Like, like-
00:30:11 Nathan Lambert: Like, like- ... did, did you identify this through Mini and Nano, or, like, howâd you come to think that this was so important?
00:30:19 Lucas Atkins: Data in general or, or just-
00:30:20 Nathan Lambert: Or like this in form of post-training
00:30:21 Lucas Atkins: ... of optimizing it for the post-training? Um, I- really ob- observing other, other players, I think. I mean, itâs, itâs... You know, the, the true base model has kinda stopped really being a thing.... around Qwen2, but definitely around Qwen 2.5, um, where you started to see how much post-training data was making its way into the, the, the base models themselves. Um, and then you start to see the models that have done that, how malleable they are with RL, Qwen 2.5, Qwen3 being a good example. And you start to see like, oh, yeah, like they are, uh, doing as much in the last probably thirty percent of training to make it so that when they go to do RL or post-training, theyâre gonna have a really good time. Um, you know, theyâre just complete-- theyâre way easier, way more malleable, way more performant than what you had in Llama 2 or Mistral 7B. So, um, I knew that i-in-intuitively, kind of going into this, but it wasnât until after Mini and Nano, yeah, where, where we kind of... Well, definitely 4.5B, where we were like, âYeah, we definitely need to juice our mid-training quite a bit.â
00:31:31 Nathan Lambert: Yeah, I agree. Okay, this was fun. We could- weâll probably revisit themes from this. I think that, um, I can definitely go over time and keep chatting because Iâm enjoying this. And for context, Mark and I had coffee at some point when I was at some conference in SF, and I was like: Damn straight, this is a fun bet that youâre making. So Iâm trying to recapture as much of this as you can. Um, for context, itâs like in July, which is similar to when you decided to start this model, which is when, like, Qwen Coder came out, Kimi came out, um- ... GLM 4.5 came out, and I was just, like, looking- and Llama had kind of been, like, become a meme of going away. And thatâs why I launched the Adam Project, where I was like: Come on, we need to have some people doing this. And I think that itâs, like, hard in the US because I think thereâs so much money to be made on AI. Like, the company- the big tech companies are like: âWe see it, and weâre gonna take it, so I donât need to bother with, like, caring about open models âcause we donât need it.â But from, like, an ecosystem co- perspective and a long-term tech perspective, I donât think that works very well for the country. So itâs kind of this weird middle ground of like, how do you convince people to actually build open models? I was on... Like, I have calls with people in government asking me, like, what would I actually do? So itâs, like, very hard to think about this. And I have this- and then itâs just, like, to hear that you guys are just making this bet on this is very fun to me, but itâs also, like, based on actual learning from trying to do this. So youâve been trying to train open models. I think Mark and I have both been at Hugging Face in our past, and youâre, you were trying to sell people on using open models, and there is a market for this, but it wasnât enough to not have the base models. So I think, like, talking about your experience in selling on-prem open models and why you needed to train your own end-to-end, and why you needed to train bigger, is great because I hope there are more stories like this, and it kind of fills a void and inspires people to work in it. So how- however you want to take this prompt.
00:33:24 Mark McQuade: Yeah, I can jump in. Um, I mean, yeah, I mean, wh- when I started Arcee in 2023, right, uh, it was... All we did was post-training. Uh, and we worked with, uh, a lot of large organizations and did model customization, you know, for their use case on their data. Um, and we were using Llama-based models, Mistral-based models, and then, you know, some Qwen. I donât even know if we actually did much Qwen, right, Lukas, at that time, but-
00:33:54 Lucas Atkins: No, we did. Yeah, we, we- Later on, but and then-
00:33:56 Mark McQuade: Later on, right? Uh-
00:33:57 Lucas Atkins: We did, and then we ended up not, because after a lot of Chinese models started to come out, then the companies didnât wanna use Chinese models, so then we kind of went... Yeah, it was kind of just tricky.
00:34:08 Mark McQuade: Yeah, and people donât realize that thatâs real.
00:34:10 Nathan Lambert: People donât realize that that actually happened.
00:34:13 Mark McQuade: Yeah, no, thatâs, thatâs a real thing. Thatâs why we, we started going down to pre-training was because, well, you know, Meta did their thing and kind of got out of it, right? So there was the, the main US player got out of it, and, and we were working with a lot of US-based enterprises that were not comfortable using Chinese-based architectures. And if you wanted to use the best open models of the day, it started to really trend towards, you know, the Chinese labs. Um, and to the point where we are now, where itâs like, you know, ninety-plus percent of the top mo- open models are coming out of China, um-
00:34:47 Nathan Lambert: Yeah, like, Cursorâs building on it and stuff. Like, people are building on these things.
00:34:52 Mark McQuade: Yeah. So, um, we said, âOkay, letâs...â Instead of we were so reliant on the Metas of the world, the Mistrals of the world, and Mistral largely stopped open sourcing, uh, you know, fully. So we said: You know what? Weâll just go down the stack, and we feel weâre capable enough to, to, to train our own models from scratch, and then we control the, you know, the stack. We can, you know, we, we control the core of, of... as opposed to relying on others to release great models. And, um, and then during this time, you know, it just happened to be that, um, you know, there wasnât a tremendous amount of US companies doing it. So, um, from our perspective, it was kind of a, a win-win, in that we were able to own more of the stack by going down to pre-training and creating our own models, as well as we were entering into a, like, a space that there wasnât a tremendous amount of competition, to be honest. Um, and, you know, I-- Lukas and I had said this yesterday, I, you know, I think as a startup, every startup doesnât want to directly compete with, you know, X or OpenAI, or Anthropic, or Google because they have more money than God, and they can do whatever they want. Um, but when youâre doing open weights, you donât-- itâs, itâs a different kind of compe- they, they donât sit in there, right? Youâre kind of going into your own path, where there isnât a tremendous amount of players, and you can kind of find your, your way and, and build your niche and, and kind of go from there and, and become something big. So, um, it kind of happened to all coincide for us back in, in July, and, and we went all in.
00:36:23 Nathan Lambert: Yeah, yeah, like, uh, the, the all-in thing is real because this is expensive. I think that- ... I could dig up in my research the cost of daily, um, twenty-four T8 B300. So I think Iâve seen this type of cost at AI too, where we have long rentals, and weâre like: I know exactly how much this costs, and itâs like, itâs not cheap. Are you... A, a way to transition this is like-... do you see the demand? Like, you were selling open models, like, does this kind of be continuous, where people are like: âYou helped us deploy this model, but itâs not good enough.â Like, is, is that something thatâs happening, and youâre like: âWell, we have this, and we can help you do it coming in this time?â Or is it like you need to build it... Itâs like, is it a we will build it, and they will come type of situation? Like, how much- ... continuity is there in this?
00:37:17 Mark McQuade: Yeah, I think itâs largely-
00:37:19 Nathan Lambert: I-
00:37:19 Mark McQuade: I, uh, from my perspective, I think itâs largely if you build it, they will come. Because we stopped, you know, focusing on that whole revenue generation side of the house when we started to go all in on being this, you know, frontier lab in the open source side. So, um, thereâs a couple pieces to that, that, that I think we should all be very proud of inside of Arcee, is that we not only went all in by committing a significant amount of capital. Like, we, we committed, you know, sixty-five, seventy percent of our capital to these models, which is a large amount for a startup. I mean, we didnât... So thatâs not like a dip your toe in, thatâs like, weâre all the way in.
00:37:55 Nathan Lambert: Yep.
00:37:55 Mark McQuade: Um, but we did that at the same time as abandoning essentially the whole revenue angle to go all in on it, because we couldnât focus on both. So we said, âWe know how to make revenue on open models. Weâve been doing it for two years. Now, letâs take a step back, because it wasnât, uh, in a repeatable or sustainable way that we h- the way we had that business set up. Letâs take a step back, letâs build these models from scratch, letâs come up with the, the Trinity family, then letâs go back to generating the revenue side of the house and the monetization piece,â which I think we are in a good position to capitalize on even more now, but we, we took a... We, we, we kind of walked away from it to do what weâre doing here.
00:38:36 Nathan Lambert: Yeah, I love this.
00:38:36 Lucas Atkins: Yeah, I mean, when you have... When thereâs only, like, thirteen, you know, uh, researchers who would... Well, weâre, weâre doing obviously our own products and own models, but when youâre working with customers, like, inevitably, those are the same people that need to help train those models for customers, and we got to a point where we were really beginning to, like, do mini and nano. We were getting down to, like, the start date of the cluster, where, um, having myself or Mark, or even, you know, Varun and others, like, pulled into customer or, or, or, uh, conversations or contracts, like, it was not-- we would not be where we are if we had continued, you have know, working with, you know, ten customers at once. So-
00:39:19 Nathan Lambert: But-
00:39:19 Lucas Atkins: ... we, we scaled that down pretty drastically. I do think that when... You know, Mark and I put a lot of thought into, âOkay, well, weâre gonna spend all this money to train these models, like, you know, w- how do we not...â I think, uh, one of the things that makes the idea of, of going all in on training open weight models hard, is that youâve seen other people try it. And, and like M-
00:39:42 Nathan Lambert: Um, like, like do you think Meta or do you think Meta or Mistral went all in?
00:39:46 Lucas Atkins: I, I think, well-
00:39:48 Nathan Lambert: Meta obviously did.
00:39:48 Lucas Atkins: I think they, they both... Yeah. I think, I think that when I say all in, I mean more like Mistral was, was one of the core ones Iâm thinking of, where- ... they were a venture-backed company that, like, had a, a, a fiduciary responsibility to bring in money, but were also trying to release open weight models, uh, for, you know, the West, and for their communities, and for the world. And, um, they tried doing closed versions, and then monetizing off of that. They, they also kind of have more recently, luckily, for all of us, gotten back to their kind of Apache 2.0 roots, and-
00:40:30 Nathan Lambert: Oh, my God.
00:40:30 Lucas Atkins: And-
00:40:30 Nathan Lambert: Have you seen the download numbers on Mistral 3 Large?
00:40:33 Lucas Atkins: I havenât. No, what is it?
00:40:35 Nathan Lambert: Oh, s- no bueno, sir.
00:40:38 Lucas Atkins: Hey.
00:40:39 Nathan Lambert: Carrying on. Sorry.
00:40:41 Lucas Atkins: But, I mean, yeah, you know-
00:40:42 Nathan Lambert: Um, Mist- the, the Large Instruct model has downloads in the last month. I honestly donât know whatâs going on. Maybe thereâs some, like, quantized version out there. I, I was confused.
00:40:50 Lucas Atkins: Maybe. Well, I mean, yeah. But I think that we-
00:40:52 Nathan Lambert: Itâs, itâs hard to get adoption. The competition is insane.
00:40:55 Lucas Atkins: Hmm. Well, thatâs, thatâs- ... yeah, I mean, and that could be a whole conversation also, is, like, how do you actually get people to use it?
00:41:00 Nathan Lambert: I was gonna ask you, like, how do you get people... How do you get people to- - really sell into this? You said youâre good at it.
00:41:06 Lucas Atkins: Yeah, I think that the-
00:41:08 Nathan Lambert: Continue your point, we can come back to it.
00:41:11 Lucas Atkins: No, no, but they... I think they all kind of tie into it, is, is... We knew that the, the market was there for, for custom models. It was two years ago, frankly, and itâs even more so now, because RL has drastically, uh, increased the areas that you can hill climb and become really powerful with a tiny model. Um, and but, but also, people are beginning to see how powerful, you know, uh, te- uh, cust- or, or training in a, a, a product is. Like, you see Claude Code, you see Codex, you see, um... I think Deep Research was kind of one of the first ones that really kind of opened my eyes to what was possible, when you kind of are kind of training in the same environment that youâre serving your users. So we knew that, that people wanted it. Weâd, weâd had good success with, with customers in the past using other peopleâs open models. So, um, it was less of a question of, like, could we monetize it, or will we? And it was just a matter of, um, could we get a model, you know, that pe- that, that we would feel that, you know, given a, a wide suite of basically being able to pick any model in the world, would, would our researchers and, and would our teams re- reach towards our own? And, uh, luckily, I think weâre there. Um, on, on the-
00:42:31 Nathan Lambert: Uh
00:42:31 Lucas Atkins: ... on the topic of, like, how do you get people to use it? How do you get adoption? You know, Iâve never wanted Trinity, uh, or our biggest advertising thing to be, like, US. You know-
00:42:45 Nathan Lambert: Yeah, I know
00:42:45 Lucas Atkins: ... like, if, if your entire-
00:42:47 Nathan Lambert: I know, man, it hurts me.
00:42:48 Lucas Atkins: Yeah, if your-
00:42:48 Nathan Lambert: I spent months reckoning with this.
00:42:50 Lucas Atkins: Yeah. If, if your entire, uh, you know, value prop is that youâre an American company-... great, but ultimately people are gonna use the best. Um, and so I think that weâre gonna be able to serve and, and the people like that need a US-based model because their compliance or legal teams wonât let them use something out of China, itâs gonna be a fantastic option. But I think, you know, kind of the next phase of what weâre doing as a company is, all right, now weâve, weâve proved to ourselves and maybe the, the wider industry that like we deserve to be in the conversation, and we can train models of this scale. Um, then itâs like, okay, how do we train the best one? Uh, âcause really, I mean, peopleâs loyalties are very fickle, and, and, yeah, you, you go to whatâs the best. I guess itâs like, how much do you think
00:43:41 Nathan Lambert: youâve learned about being able to tune a model narrowly by going and building the whole stack? Um, something we talk about is like ability- ... to specialize models, and I kind of, of opinion that you just make a better general model right now âcause the pace of progress is so high. And but the question is like, can we tune a OLMO thatâs very good at science or something? And I- ... w-would guess that training the entire model, youâre going to be able to actually do a better job at what you were doing, but I donât know how to articulate why or what that looks like.
00:44:18 Lucas Atkins: Um, I mean, the, the, the simplest answer to that being yes is just that... or the simplest reason why thatâs the answer to the question is yes, is because we know what went into the model. Like, we know what it actually saw at the later stages of training during the decay. Um, and so that all- that helps influence, A, what are we tr- what kind of data and what topics and, and what format are we giving these models, uh, in post-training? But it also allows you to know like, okay, where, where do I absolutely wanna crank, you know, how, how many- how much of this, say, 230 billion dataset, do we want it to be math or, or, or, or coding? And a lot of thatâs influenced by what youâre able to put in-
00:45:06 Nathan Lambert: How, how much of your post-training-
00:45:07 Lucas Atkins: ... post-training
00:45:07 Nathan Lambert: -do you expect to redo? Like, uh, how much can you say about when youâre serving something on-prem? Um, you- youâre not gonna redo the pre-training. You might, for a very big customer, redo mid-training or do continued pre-training- ... in which, in that case, you do need the pre-training data to keep, keep it being stable. Which is a use case where like Iâm- I would love to see a paper thatâs like, âBecause of OLMO being open, we continued to pre-train on biology, and we mixed half of their exact mid-training dataset in with our dataset, and it, and it worked,â yadi, yadi. Like, you could obviously- ... do that, but how much do you think is gonna be like the standard, you fine-tune the last instruct model, or do- are you gonna have to retouch the post-training for a customer? Because that, like, I, I really feel like-
00:45:48 Lucas Atkins: Um
00:45:48 Nathan Lambert: ... itâs just at the end.
00:45:50 Lucas Atkins: It, I think, I think-
00:45:50 Nathan Lambert: But it would be fun if you had to change it.
00:45:52 Lucas Atkins: For the most part, um, I think a lot of tasks will be fine just starting from our, our, our, po- uh, like the released, you know, official post-trained version. Um, now, thatâs for maybe simpler tasks, is the wrong way to frame it, but if itâs like, âOh, hey, weâre doing a deep search agent. We want it to do 30 calls and, before...â That would be a good use for just starting with the finished model that we released thatâs already post-trained. Now, if weâre going into something along the lines of, um, a very low-resource programming language or, um, something that it didnât see a lot of in, in, in pre-training, um, or itâs kind of like a, you know, weâre wanting to train this thing to be really good at humanities last exam, but tools. Um, once we get into the world where weâre having to, especially... Actually, I have a much better answer to this question as I was thinking through it, but most of that holds the same. I think that the, the, the world where weâre gonna be doing a lot of extra instruct and, and SFT and, and post-training is gonna be when weâre trying to distill capabilities from large, like into mini or nano. So say like, oh, you know, this large is, is, is really great at invoice processing, but itâs also 400b, and the, you know, the company doesnât wanna be hosting that on-prem, you know-
00:47:24 Nathan Lambert: Ah
00:47:24 Lucas Atkins: ... letâs go out generate a new one.
00:47:25 Nathan Lambert: Do you have costs off the top of your head for, like, what the hosting costs are for each of the model? Like, do people... Are people all gonna host these models in the same way, or is there actually-
00:47:32 Lucas Atkins: Uh
00:47:32 Nathan Lambert: ... a wide variance? And if you have, like, the same three models- ... do almost all of your customers end up hosting the same way, or do you end up doing a lot of, like, how do you configure the model to fit in the right hosting for them? Like, is that part of-
00:47:44 Lucas Atkins: It depends
00:47:44 Nathan Lambert: ... the business model?
00:47:45 Lucas Atkins: It, it, it, it kind of... And we tried to move a, a, a little bit further away from that because you get into the risk of being like, like a consultancy, and itâs- that becomes tricky, where thereâs not a very clear separation of concern. But, um, for the mo- it would change depending on, were they using AWS? Did they have a commit with Azure? Um, if not, okay, then we, we can go to, you know, someone like Prime Intellect or Parasail and, and get a, you know, maybe a, a cheaper rack of eight. Uh, it just really depended. Uh, thereâs quite a bit, um, of, of people that were also serving them, just using, like, Llama CPP. So, like, on CPU-
00:48:25 Nathan Lambert: Uh, is the 400b designed to be, to fit onto one rack of eight 80 big gigabytes in FP8? Is that how you designed it? âCause Llama- ... Llama four, whatever, Llama 405b was the same. It was like one rack in FP8 works pretty well.
00:48:41 Lucas Atkins: Itâll do- we... well, youâll be able to get really good throughput, a little bit lower concurrency on a, a rack of eight H100s at FP8, and then for, like, our, you know, what weâre serving, weâre serving them on, uh, a series of H200s, but weâre not doing, like, multi-node inference. Uh, but thatâs just to add more, you know, replicas and- ... other kinds of things.
00:49:03 Nathan Lambert: Hopefully, eventually. I think that the-... Do you have anything else to say about selling open models? I think that generally, like, how do you think about the market for AI? âCause I see the market as being so big, but the- with specifically with open models, itâs so hard to measure. I think Iâve started talking to some of the Chinese labs at all- as well, and I like to ask them, like, this is very US-centric and like Fortune 500 or whatever, and itâs just like, who the heck uses these models? I think- I guess another question is, like, what license or do you know the licenses youâre gonna use for the biggest models? And I think theyâre, like, youâre, youâre playing with fire âcause people can use it for free, obviously, but potentially- ... youâll get to hear like, âOh, s**t, somebody actually used our model for this.â And I think any successful business, youâre gonna want... You, you, you know that this model is not gonna be very relevant in a year with the pace of progress. So like- ... how do you think about your license decisions?
00:49:55 Lucas Atkins: Uh, we- you know, with the 4.5B, we tried to do like a, like a, a reve- one of those revenue-gated licensing. So itâs like, oh, itâs completely free for you to use for commercial and whatnot, but if you or your company made over, I think it was like $1.7 million last year, then you need to come to us and get a license. And what we ultimately found was like, it, it didnât... Maybe for some people who are just only trying to train the model, release it on Hugging Face, and then just call it a day, maybe that is a huge requirement. But when so much of our, our, our company is built around, you know, training custom versions of the models, and, and not even just ours, but in general, even before we did pre-training. Like, at the end of the day, i- as long as we were using it, a- and we knew that we were in full control of, of whether- if we really succeed, itâs because we trained the models, we did them well, and we executed on it well. If we fail, itâs because we, uh, didnât execute, instead of, oh, some company just stopped releasing good open models. Um, so we eventually switched to just Apache 2.0, and Trinity Large is also gonna be Apache 2.0. Um, you know, Iâm- I think it is-
00:51:23 Nathan Lambert: I think this is the right approach. I have a big investor-
00:51:25 Lucas Atkins: Yeah, I think it-
00:51:25 Nathan Lambert: Without, without naming other companies, itâs easy- like, raising a lot of money, whe- or being Meta and releasing open models, and do it- and you could release it with non-commercial, and you could get all these, like... You could talk to, I donât know, f*****g Adobe, whoever. Oh, Adobeâs too big. Theyâll have good AI. Some... I donât know, a bank. Bank of America. You could run Llama on Bank of America and make good money on this. But I just feel like the cultural home of open source AI, and I donât think- itâs impossible to know who wins it, and I donât think that youâre in the prime position, and I donât think that itâs easy to win, but youâre doing a thing that aligns with it. Itâs the person that just, like, commits to building the models and learning how the ecosystem works, and to rebuild the models based on the feedback th- that you get from people, and to just kind of commit to an evolving process. And if the whole thing works out, there will be a lot of value, and the person who understands it best should be able to learn how to extract said value. And I think that Iâm personally, like, sometimes frustrated with Hugging Face, âcause I feel like they have sat on that s- a sort of position like this, and they- ... havenât figured it out. Not that it is easy to figure it out, but I think that has to be the ideal of open source AI, of like, if itâs really gonna work, thatâs, thatâs what I hope it looks like. And itâs like, I, I donâçź know, maybe you guys could do some of that. Like, I have a question of like, could you figure out how to make models that are more fine-tunable- ... after all this post-training? Because you need to sell it to a- you need- ... you, you know the customerâs not gonna want it off the shelf. And I donât know how to train to post-training to make sure that you donât, you donât cook it. Maybe you just learn that you need to warm up the model in a l- in the right way, and you just learn the technique of training downstream. But when you talk to people doing research, the different base models have such different characteristics. I think one of them is character training. I did this paper, and the guy was like: âQwen and OLMo love their character,â and Iâm like, âI have no idea why.â And but itâs like Llama and Gemma, you can change them so much. And Iâm like, âDog, like, please figure out why this is the case.â And for one thing, itâs really cool, but also, like, in your case, that would unlock a lot of value to be like, we know exactly what the modelâs gonna do, and we know exactly how to change it. So.
00:53:35 Lucas Atkins: Yeah-
00:53:36 Nathan Lambert: Uh
00:53:36 Lucas Atkins: ... it, it, thatâs- no, youâre, youâre, youâre right on the money. I think that even, uh, going into the post-training at large, we, uh, one of our researchers came out with, like, a pretty cool, um, experiment and ablation run that they did on drastically reducing catastrophic forgetting. And I almo- I mean, this was, like, three days before we were gonna start doing SFT, and then we ultimately just... I, I ended up pausing on it because it was just throwing something in that wasnât tested. But, um, yeah, I think-
00:54:08 Nathan Lambert: A good research lead. You did the right thing.
00:54:10 Lucas Atkins: Yeah, I think, I think one of the most important things long term, you know, as we look at kind of what our research priorities are for this year is, is thereâs obviously just how to scale RL and, and make these- the end result of the model as good in as many situations as possible. Um, but I think the other half of that is, you know, how do we make the, the, the speed and efficiency and, and performance of customizing them as, as fast as possible, and as easy as possible.
00:54:42 Nathan Lambert: Yeah. Do you learn in making open models from your experience just kind of running these open software things in MergeKit and DistillKit? I know there was a whole license journey on one of those as well.
00:54:52 Lucas Atkins: Yeah, DistillKit.
00:54:52 Nathan Lambert: Do you feel like theyâre kind of isolated?
00:54:54 Lucas Atkins: Or MergeKit. Um, yeah, I mean, I think so. I think that, that, um, you kind of have to play the tape out. With MergeKit-... it was by far our most popular piece of software weâd ever released, but it was so popular because it took something that isnât fundamentally very complicated, but we ma- but itâs time-consuming, and standardization is great for things like that, and we made it, uh, you know, streamlined and easy to do and fast, and you could experiment and ablate really quickly for, you know. And, and so I, I think that when we switched that to, like, a, you know, a, a similar, uh, revenue-based licensing, like, it, it didnât end up having the value prop that was important because are you gonna pay Arcee, you know, thousands of dollars, or are you just gonna have one of your researchers-
00:55:52 Nathan Lambert: Youâre gonna have clone code in a week, right?
00:55:52 Lucas Atkins: recreate it in a week, right? Yeah, so itâs-
00:55:55 Nathan Lambert: In a day.
00:55:55 Lucas Atkins: Itâs, itâs kind of... It, itâs remi- itâs remembering like, okay, what is- what problem is this solving, and is this even a prob... Like, is the solution to this monetizable? Um, and so MergeGit, we brought it back to the original license, but I think with even viewing the models in the same way, itâs like itâs... Open source is an unbelievable marketing tactic. Like, thereâs no one would care about Arcee if we werenât open sourcing stuff, âcause as soon as you do something closed source, if youâre not the best or the cheapest for your price point, I mean, your performance point, no oneâs gonna use it. Because-
00:56:30 Nathan Lambert: Um, another question on this. Um, do you think that open models are kind of at a disadvantage when progress is so high? Because itâs potentially easier to swap APIs than open model configurations, especially if, like, model weights are changing sizes or something like this. Where itâs like, âOh, I can just upgrade to the new Opus, and I do this.â Like, does that, like, uh, decentivize people from using it? Or do you think most of the people are like: âI can only use open models, therefore, Iâm gonna use open models?â
00:56:56 Lucas Atkins: Uh, I think for the people who are using, like, s- either self-hosted or, you know, um, uh, bespoke, uh, you know, engines to, to run it, where they have complete... You know, in a VPC or they have complete control over, like, data in and out, egress, ingress. I donât think thatâs really gonna be so much of a problem because theyâre obviously doing it for a reason. Um, like, theyâre either for privacy or security or, or HIPAA or SOC 2. For whatever reason theyâre doing it, um, I, I donât think that thatâll be, um, so much of a blocker, but I definitely do think that, um, you know, by far, e- even, even with some of the, the larger open... You know, like inference players, like Together and Fireworks, that, that host a lot of open models. Like, being feature- being on feature parity with a lot of these, these larger labsâ APIs is gonna be extremely important, um, o- of being able to serve, you know, um, with features that theyâre used to, like prompt caching, that kind of stuff.
00:58:03 Nathan Lambert: Yeah, are- like, I, I think I saw that you guys are setting up an API as well. Is that kind of what the vision there is, is being able to o- offer parity at least, or, like, make it easy for people to consider it?
00:58:13 Lucas Atkins: I think so. I, I- weâre- we very... Yeah, we are doing our own API. We are hosting it. Um, we havenât- we, we push a lot of that through Open Router just because itâs such a great place to get, like, discovered. Um, as... If we see, like, tremendous growth there, that would obviously be where weâll, weâll invest very heavily. Um, whereas the right move might be to let other people host it, and we invest super hard on the infra for, like, make- taking advantage of the models, um, and, and customizing them. Thereâs, thereâs, thereâs a few avenues we have ahead of us then, and we have, you know, projects going kind of toward to poke at each one. Um, and weâre just kinda getting as much data as we can before we... I mean, weâre gonna have to go all in on another direction soon. Not, not like pivoting away from pre-training, but now that weâve done that, now w- whatâs the next big bet weâre gonna make, and how do we go fully into that? So weâre trying to figure out what that is.
00:59:12 Nathan Lambert: Yeah. My two last kind of, like, real questions are, like, one is... I guess I can start with, like, where do you see the open model ecosystem? Do you think- where would you see it changing substantially in the next six or twelve months? I, like... Or, or do you? Or you just kinda think weâre marching along for a while?
00:59:31 Lucas Atkins: No, I think weâll, I think weâll, weâll be... I, I, I donât think itâs an unrealistic prediction to make that by the end of 2026, like, the best model in the world is, is some degree of open. Uh, I think thatâs very, very possible, especially with, like, what Iâve seen GLM and, and MiniMax do recently. Um, they have started to find that secret sauce that takes you out of just being good on benchmarks and, like, genuinely useful in peopleâs day-to-day workflows. And, um, I wouldnât- like, if, if I, you know, came back, and I... Someone came from the future and told me that the best model in the world was, uh, an open-weight model, I wouldnât be surprised. I actually think weâre on a, a, a super good trajectory, and, and, and fostering and, and promoting that kind of work and adoption here in the United States is gonna be extremely important.
01:00:24 Nathan Lambert: And where do you see the company going? âCause like, like, I have my guess. Like, you kind of hopefully-
01:00:31 Mark McQuade: Whatâs, whatâs your guess? I wanna hear your guess.
01:00:31 Nathan Lambert: Um, you can hopefully do a mix and kind of oscillate into trading when you get... Like, you need to start having the feedback of the real world. I think thatâs obvious. Like, itâs o- like, itâs... Well, obviously, you need to make money to survive as a company, but then you need to start using that as the feedback to guide training. And then itâs like, you need to figure out how to balance and do some of them at each time, and you can plan your cluster at different times, and then you kind of... Hopefully, they become a, a loop across each other, and they kind of make it so obvious of why you each need them, âcause it, it seems somewhat natural.
01:01:03 Mark McQuade: Yeah, I mean, exactly. You know, you kinda hit, hit it right on the head. Um, you know, getting feedback and then kinda steering the ship from there, um, is, is probably-
01:01:15 Lucas Atkins: ... exactly what weâll do, but we have a good idea already. I mean, first and foremost, you know, we talked about it earlier, w- weâve spent a tremendous amount of money. So, uh, we need to go raise some money after we - after we get, you know... We need people to back the, the, the mission and the vision of US open source and, and, you know, so, um, because, uh, you know, we, i- i- Lucas had mentioned about, like, MergeKit and how we flopped the license and, you know. I mean, weâre a smaller-sized start-up. We have-- weâre-- we gotta think of kinda unique ways to try and generate revenue because we donât have the money of the large labs. So, uh-
01:01:52 Nathan Lambert: Well, I think itâs a benefit to the employee. I think a lot of these labs have over-raised.
01:01:56 Lucas Atkins: Yeah, I like, uh- uh, I-
01:01:57 Nathan Lambert: OpenAI, Anthropic, and all of them are fine. Like, with the OpenAI, Anthropic, Cursor scale, like, let it rip. They should, they should really rip the raising. But all the other companies that are stuck at the, like, the one to two billion range without, like, obvious traction, like, the risk goes to the... I mean, you could-- a lot of them do secondary, so a lot of the founders get out. But itâs like, the risk is the employees get nothing.
01:02:21 Lucas Atkins: Yeah. Yeah.
01:02:22 Nathan Lambert: There is a lot of money, but thatâs also why I like the approach, âcause itâs like, âOh, youâre doing the actual start-up thing.â
01:02:28 Lucas Atkins: Yeah, yeah. Yeah, I mean, I think... W- what I was gonna add to what Mark... is just like, what- whatever we do from, uh, uh, uh, scaling and, and speeding things up and growing, um, my goal is to keep our research and engineering teams pretty small. I think, I think that one of the reasons weâve been able to, to move as quickly as we have is itâs been, like, a small group of, like, highly intelligent, smart, and opinionated people sitting in a room, debating in good faith on decisions. And I think that thatâs, uh, uh, under the constraints of, âHey, we donât have five hundred million dollars to go and, you know, to rip on, on, you know, X, Y, and Z.â So and I think thatâs kind of where creativity comes from, and I think that fostering a culture like that over time is how you can kind of make it so that excellence is less of like a, um, an accident, and itâs actually, like, a by-product of the way that you work. So, so weâre gonna stay small, weâre gonna stay lean, but, um, I, I do think that, like, the, the major, um, kind of challenge for us over the next probably six months, beyond any other models we might have, kind of, uh, think or weâre thinking about, is, is getting up to, like, post-training parity with the likes of DeepSeek, and GLM, Qwen, and others.
01:03:47 Nathan Lambert: Yeah. I, I hear lots of horror stories about this, where itâs usually and-- itâs-- you end up having people that are going after different important abilities, but, uh, like, doing each of the abilities alone is pretty easy to hill climb, but then you just end up with such a mess. Itâs like youâre- ... building a custom puzzle, and youâre building all these custom pieces, and theyâre magnificent, and then youâd have to, like, pick up these pieces and assemble this unknown thing at the end. And itâs like-
01:04:12 Lucas Atkins: Like they didnât have the same designer, right? Yeah.
01:04:15 Nathan Lambert: As AI2 is barely scratching the surface of this. Like, you talk to the people at the frontier labs, and itâs like, holy cow, like, post-training is really the Wild West. But a lot of it works. I think, like, we find-- like, even like model merging gives a ton of performance across the whole- ... training pipeline. Itâs like- ... you merge at pre-- you merge after each pre-training stage, you merge in post-training. Itâs like-
01:04:35 Lucas Atkins: Roon can tell you.
01:04:36 Nathan Lambert: But merging post-training becomes a lot more complicated because you- ... can have all these domains and things, uh.
01:04:41 Lucas Atkins: Well, in, in merging, you know, it, it actually, it used to be very YOLO, um, the way we used to do it, and, and Charles, who, who created MergeKit, I call him, like, chief alchemist, and, like, youâd kinda just send him ten promising checkpoints, and heâd come back a day later with, like, some insane, you know, model that was really good at all of them. And, and you canât do that as much in post-training anymore because of, uh, of just the, the formatting and the way that RL is done. Like, you do have to be a little bit more surgical about it, but yeah, everyone can tell you, like, any time we start to see anything worrisome at all in training or, or, or even something going really good, you know, âLucas, what do we do?â Iâm like: Merge it. Iâm like, just-
01:05:21 Nathan Lambert: Merge.
01:05:21 Lucas Atkins: ... Iâm like: âJust take it, just merge it. Letâs see.â And more often than not, it fixes it, so...
01:05:27 Nathan Lambert: Um, do you merge during RL? Like, you could just, like, merge the last few checkpoints and resume or something?
01:05:32 Lucas Atkins: Weâve ex-- weâve, weâve dabbled in that, not, not for what weâve done. You know, again, a, a lot of the, the mini, nano, and large story for Trinity is, like, getting to a level of... what was my level of complexity I was comfortable with us undertaking, and then, uh, not introducing anything more. So, um, not yet. But we, I mean, we, we, uh, regularly merged. We didnât do it for LARP, but we used to merge a lot, um, during just, like, your standard, uh, um... When weâd do, like, RLHF, we used to do a bunch of merging. Weâd do it, like, every five checkpoints. We would-
01:06:11 Nathan Lambert: Online RLHF or D-DPO?
01:06:13 Lucas Atkins: Thereâs DPO.
01:06:15 Nathan Lambert: Yeah. Itâs so much easier to get started. One of my goals is to have somebody figure out how to do actual online RLHF, pure LM feedback, obviously, for scaling. But itâs just like- ... itâs, itâs unsavory to itâs just, like, doesnât look like DPO-
01:06:28 Lucas Atkins: Yeah, I mean, if, if, you know, if GRPO and kind of op-- in, in the, the present day RL regime, like, if that hadnât materialized when it did, I think that wouldâve been a big topic in 2025. But I do think that, you know, GRPO and just the overall, um, DeepSeek and o1 style reasoning and thinking and RL kind of... Any, a- any person who is thinking of doing that for, like, performance reasons, realize that there was something that had fifty thousand papers released every day on how to do it. Um- ... that was kind of probably right where youâd get the same amount of performance.
01:07:07 Nathan Lambert: Um, do you force dog feeding? Do you make yourself-- do you guys use your own models to understand them? Like, do you, like, make that a thing?
01:07:14 Lucas Atkins: Uh, Mini was the first one we could actually start doing that with, um, a- at least for, uh, a more general day-to-day tasks. So a lot of our, like, internal Slack, we have stuff that, like, monitors Twitter and LinkedIn for feedback on Trinity and, and, and that kind of stuff. That all runs on Trinity Mini now. Um, and then, uh-... you know, we, we put a good amount of work into, into large being, um, you know, good in, in a bunch of your, like, OpenCode and, and Cline, uh, and, and Kilo Code. So, um-
01:07:45 Nathan Lambert: Uh, what does that, what does that work look like?
01:07:49 Lucas Atkins: Uh, working with those guys to get data. And then, um-
01:07:53 Nathan Lambert: Thatâs, I mean- Good for me to know.
01:07:55 Lucas Atkins: I mean-
01:07:55 Nathan Lambert: I should do that, I guess.
01:07:58 Lucas Atkins: Yeah. Yeah, working with, uh... Or, or I mean, it- the way it started was us, like, using open models and then, like, passing those through as the base URL, and then, like, getting the logs from that. Um, and then realizing that, like, that translated pretty well. Um, and then over time, obviously turning this-
01:08:16 Nathan Lambert: Um, can you expand on this? So I was gonna ask you-
01:08:19 Lucas Atkins: So-
01:08:19 Nathan Lambert: -if youâre, like, using these open models regularly, âcause I, Iâm just, like, Claude Code psychosis, man. Iâm like, âCanât take that away from me.â
01:08:26 Lucas Atkins: Yeah, I, I use, I use four... Iâve used 4.7 a lot. I think 4.7 from GLM was one of the first ones that could replace a lot of my day-to-day. Uh, Iâll still reach for Claude Code or even 5.2 Pro if itâs, if itâs, like, something thatâs, like, really... I- if I do not know how to measure what success looks like for something, Iâll usually use those. Um, but, uh, yeah, I mean, it, it- even using DeepSeek before, um, kind of their May update was hit or miss. But, um, yeah, w- the reason I decided to, like, start talking to these people and working on, like, how can we get data and, and start making our models good in these systems was I would use them. I had a, um, you know, something that would grab the logs, like, it, you know, inter- as a proxy, so itâd like grab the logs and then format them in the messages format. And then I saw that and went, âYeah, thatâs... You can make a pretty good filter for just, like, standard stuff that you donât want, and kind of hit a scale.â
01:09:30 Nathan Lambert: Yeah, it makes sense. So, so youâre like, uh, open code will let you look at the data, and then youâre probably gonna get a sense for... Like, I donât even actually know how the, on the back end, the code agents in open code format data, which I think is actually something I should just go look at, âcause then you can design around.
01:09:44 Lucas Atkins: Uh, theyâre all different. Yeah. Yeah, but you just have to- you just- basically, it all starts from like, what do you want your format to be? And then how can you take what, what those look like to, you know, to... How do you force it into that? The hard thing, though, is, is with newer models like MiniMax and 4.7, the way they do interleaved thinking is, is like... You know, Iâm a big believer in post-training. Like, if youâre gonna do interleaved thinking, like, every sample in your data set should be that. Um, it, you know, it should follow that same format and that same behavior. So, um, that gets tricky if youâre trying to, like, take a bunch of Nemo tr... Or, or, or, well, like, uh, DeepSeek data and Qwen data, and then, oh, weâre also trying to mix in MiniMax, and at that point, youâre- it, it gets really difficult âcause they all handle thinking slightly differently.
01:10:34 Nathan Lambert: Yeah, I can buy this. Um, okay, this was fun. Any last predictions or things you want people to know about the model? I will say that, um, when you debuted the Trinity models, you had a great blog post that was very to the point, that covered a lot of this. So Iâll definitely link to the, um, what is it? The Trinity manifesto. I enjoyed reading it. So Iâll link to that in the show notes, and, oh, hopefully you have a new one for me to read when youâre done with the model.
01:10:58 Lucas Atkins: Yeah, weâll do- we will have a tech report. Weâll have a tech report for you, too. So we, we never, we never did a tech report for 4.5B Mini or Nano because we were so focused on just getting to large, but we also thought itâd be very interesting to write it under the, the... How do you go from 4.5B to a 400B MoE in six months, and, like, what did we learn-
01:11:19 Nathan Lambert: Thatâs right
01:11:19 Lucas Atkins: ... when youâre viewing it as a whole, so.
01:11:21 Nathan Lambert: Thatâs about the timeframe that, um, Ant Ling took, too, as well. Ant Ling, uh, the anchor, we talked about, theyâre like... It took us about six months to do, um, Ring-1T and their 1T models, which, like, it sounds like a lot more, but I think thatâs about the same. It, it depends on compute and configs and stuff to go from, like- ... basic modeling to big MoE, which is pretty interesting to see a lot of people speedrun this sort of thing.
01:11:46 Lucas Atkins: Yeah, itâs, itâs a really, uh... It is a logistical nightmare, but, like, I think everyone on the team has had a tremendous amount of fun over the last, uh, six months. So now the fun begins.
01:11:58 Nathan Lambert: Yeah. Congrats on the milestone. Congrats on the model existing. That has gotta be an almighty relief, and Iâll look forward- ... to see what you all are up to soon. Iâll stop by at some point next time Iâm in the Bay.
01:12:10 Lucas Atkins: Yeah. Yeah, come by. Yeah, come by.
01:12:12 Nathan Lambert: Thanks for-
01:12:12 Lucas Atkins: Thanks for having us.
01:12:14 Nathan Lambert: Yeah. Thanks, guys.
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.interconnects.ai/subscribe - Laat meer zien
