Just realized that there are basically no American open models right now ever since the Llama series was abandoned. Basically Gemma and GPT-OSS I guess?
Ah but Mira Murati's new Inkling is Apache 2.0
But it makes sense that if you're a university researcher you are thinking about what's a model that will be open weight and developed over the long term and doesn't raise 'Chyna' concerns in Washington DC
Not only are there many American open weight models as others have mentioned, but Americans are the only ones doing actual open source models [0]. Not just distributing binary blobs and calling them "open".
There aren’t any relevant ones. I think it should be an important goal of these projects (unless there is a clear conflicting goal) to make models people actually talk about and use.
There is no question this is true for the Chinese open LLMs. GPT-OSS had a small moment of interest, arguably it was a success as an open model for a while but it’s not relevant now. The early llamas were probably the most successful for their time.
Allenai / olmo was never relevant as far as I can tell. It’s not super helpful, especially as a sovereign government initiative to build an also-ran, they should be going for real relevance.
LiquidAI LFM models are amazing, but very situational. IBM Granite series are also unique and interesting for trying to reduce liability and extend local context size. Nvidia ships some and there was also that Inkling model recently. Poolside just released theirs.
Meta might release something this year. X AI's Grok is still due to release a model, if Elon keeps to his word even if they only release a distilled version. Reflection AI has been quiet, but their access to compute is ramping up. Microsoft's MAI is considering releasing some open weight models which would be great to see!
Ilya's SSI is unlikely to release an open model since he's aiming for radical safety. That bet could pay off if the existing approach produces so much chaos within the next 10-20 years that some global ban is achieved and a super safe model is promoted as the compliant route.
We don't get many huge model releases though. I think it's harder and more expensive to safety align them. Even if you do, people will work around the safety and abuse the models. Plus it makes it even easier for Chinese companies to distill things that aren't as easy over filtered APIs.
There is a lot of internet propaganda to the effect that the US is simply unable to release open weight models or that China has so many more AI companies that the US is drowning in Chinese open weight models, but it's more like we're being careful and China doesn't care. If you host a model in China, it has to be censored and downloading any models requires you to provide your identity. Huggingface is banned there. When they release their open models in the west, they don't have to care whether the models are aligned in any way.
What? US laboratories are currently unable to contain their agents while doing security testing, and besides that, time and time again US labs seem to put short-term money above long-term safety.
Wasn't that literally why they tried to oust Altman from OpenAI, as he basically was 100% focused on profits and tried to cut down on safety across the board and lied to get his way?
> If you host a model in China, it has to be censored and downloading any models requires you to provide your identity.
I'm not disagreeing with that first part (obviously that's about inference hosting, not creating/training weights or hosting those weights), but the second part I'm not so sure about. AFAIK, ModelScope (which is the Huggingface in China) seems to allow downloads without verifying any identity and also hosts a bunch of abliterated weights.
We don't yet have US regulations and testing labs. Obviously that would be a good thing to have. I mean like the equivalent of the FCC. If you ever release a hardware product then you know what that entails.
Is it not illegal to "hack others" and "defeat protection/defensive systems" in the US already, including for both individuals and companies? Regardless if it was "by accident" or not?
It's difficult to tell if you are for or against access to open weight models as a general rule, so I am curious to hear your opinion on this.
Personally, I think we will one day come to see access to open weight models as an inalienable right to defense against tyranny, the way the second amendment is framed today. Just as encryption has become, which we similarly had to fight for in the 90s. I also understand that some regulation is sensible, but that doesn't automatically mean mandatory restricted or supervised access; any such restriction has to be extremely well-justified as essential for protecting the liberty of the people.
And as far as supervised access, whether or not identification is "handled by a third party" or "data is deleted after verification is complete" is immaterial; a citizen must not be required to trust their government. Any trust can and will be abused given enough time. Our systems must be trustless, and any expansion of government must be matched by an expansion in citizens' ability to check said government, in order to stand the test of time.
So supervised access seems completely off the table. And this can't just stop at access to models. Because linguistic analysis is a thing, and LLMs are scarily good at it (and existing non-AI solutions are still quite good given enough data), even the possibility that a government or other entity can save your messages means you've opened yourself up to deanonymization and surveillance. The chilling effect this has is undeniable, and the Supreme Court has made it clear that we cannot authorize government policy which creates chilling effects against essential liberties. Not to mention the possibilities that each category of users may be served subtly different models designed to influence them or constrain their agency/capability.
We're left with a situation where distributed access to capable open models is the only defense against a government or NGO which has access to billions of dollars of surveillance infrastructure and compute.
"I think we will one day come to see access to open weight models as an inalienable right to defense against tyranny." I don't think this kind of rhetoric about individual civil liberties is realistic any more when the next centuries belong to China, and even countries with a liberal democratic tradition are converging towards the Chinese model.
>even countries with a liberal democratic tradition are converging towards the Chinese model
What specific examples of this do you have in mind? I can’t think of any liberal democratic countries converging on a combination of (a) single party rule, (b) nearly universal intrusion of state or party actors into private sector entities, (c) financial repression of private investments, and (d) the associated suppression of domestic consumption.
I meant more generally: many countries are recognizing, just like China has, that the fundamental challenge of our modern era is ensuring social harmony. The OP's belief in individual civil liberties as a good in themselves is anachronistic now.
The thing about inalienable rights is that they are not rhetoric, they are an intrinsic recognition of rights that do not require the recognition of authority: Governments which do not respect these human rights should not be modified; not the other way around.
China is an authoritarian government and its policies have no more bearing on what people settle for than the currently socially unacceptable regime in the US.
In my opinion, if one lacks the motivation or resolve to fight for these rights, they should do so quietly and not attempt to patronize others who still stand by these rights as not being "realistic".
Allen Institute for AI has quite a range of very interesting very competent more specialized models, for earth sensing, embedded robots, for others. Their SERA model shows a remarkably capable model for such a deliberately small investment effort, with documentation on how you can train such a model yourself or refine it easily at little cost. Their EMO pioneered a better MoE with great numbers (at least at the time). https://allenai.org/
Laguna S 2.1 is really great too, in the "preview" release they've done so far at least. Still pending some reasoning-looping, but besides that, it's a really strong model to run within 96GB VRAM with the NVFP4 variants, and it's really good at coding (specifically).
Well, bit more complicated than that, I've been eagerly helping in testing and keeping track of what they've done. Initially there were serious bugs, also about the templates, eventually they released RC1 which had some fixes towards the looping. Then a couple of days later, they released RC2 which supposedly fixed the issue, but ballooned the size so all of us who were running Laguna S 2.1 on a single Pro 6000, suddenly could no longer. So, unsure if RC2 actually fixes the issue, as we're a bunch who can no longer run it :)
Besides that, it was also discovered that their suggested inference parameters were wrong and led to worse behavior. Eventually someone discovered these works best (so if you have the issue with looping right now, try these, helps a lot for me but not 100% still) and was also what the evals used apparently: temperature: 1.0, top_p: 1.0, top_k:20
Now we're waiting for RC3 which Poolside said will come at one point, and hopefully also brings down the size again NVFP4 weights + full context can load properly again even on "smaller" hardware.
I found Laguna S to be pretty good at coding, pretty fast, but pretty bad as an agent - not proactive, would frequently stubbornly argue things that weren't true, and pretty bad general knowledge.
But as a pure coding model, pretty good.
Deepseek v4 Flash 0731 is so much better if you can run it, though.
Grain of salt, I think I grabbed Laguna after they fixed the initial looping issues, didn't notice those, but there might've been other fixes since.
Yeah, this is my perspective too on Laguna S 2.1. Works amazingly for coding, pretty bad for pretty much anything else. I don't do a lot of advanced math, supposedly it's good for that too.
Like glm-5.x I think it has enormous self introspection that it often trips up on, but that this self reflection is actually a superpower, that enables incredibly good output. And from (in some cases) very small models.
If you watch it think, which you can, unlike American closed models, you can steer it. You can provide a a massive rocket ship stratospheric boost to help it orient itself. You have no self correction, there is no multiplayer in American proprietary models.
Sure it's great having super powerful mystic oracles that have the "right" answers. But I love respect & revere the open thinking. No it's not automous. But it is brilliant. And it considers. A lot. Deeply. It chases. That to me is the most human of models, even as it falls far astray.
You should help it. You can. Unlike these vicious dark surfaces which yield and tell you nothing. I think this is the actual meta-core-super-point of "The session you cannot take with you" (link below). It's the session that does not care about you, will not interact with you, will not peer with you, that is a dead remote far off oracle to you. Fuck these "oracles". They are a plague against the human spirit. We should alloy humanity and AI to Augment Intellect (Engelbart). (To do less is species treason.)
https://earendil.com/posts/session-portability/https://news.ycombinator.com/item?id=49118781
Yeah I think it got bad press because the chat templates (or something?) were messed up on first release, but I've been using a quant of it and it's a powerhouse, better than qwen 3.6 27b for local on a 3090, which is saying a lot.
No, the quants they released were also messed up. RC2 also ballooned the size so the ones who were excited about RC1 (like me) can no longer fit it in our hardware. They haven't promised anything, but said they'll try to restore the RC1 size for the next update of the weights.
Looks like on <https://arena.ai> agent arena (grouped by lab) Nvidia is 15/15 (much worse than Thinky and Mistral) and on text arena it's 18/27
On <https://openrouter.ai/models?order=most-popular> I definitely see usage though (probably mostly cause Nemotron 3 Ultra is free) the grouped order is DeepSeek, Tencent, Xiaomi, OpenAI, Z.ai, Nvidia
I think glancing at a random snapshot from today misses all the context. Nemotron 3 is far more significant than you're giving it credit for.
At this point, Nemotron 3 is really an 8 month old model series. That's when Nemotron 3 Nano was released, and the Nemotron 3 Super/Ultra models this year are obviously based on that recipe, mostly just bigger with a few tweaks here and there. Against today's models, no, not that interesting. Each of the Nemotron 3 models were briefly competitive when they launched, but never exceptional, and less competitive with each scale up. The fact that it took so long for Nemotron 3 Ultra to launch really hampered its competitiveness.
The Nemotron 3 series is extremely open about training recipes and training data, far more open than most open weight models, and that is valuable.
Before Nemotron 3, Nvidia had never released a single LLM that I would consider interesting at all, so Nemotron 3 was a big step up. The closest thing was Mistral NeMo, but a significant part of the credit there goes to the Mistral team, not Nvidia.
Given how much Nemotron 3 improved, I'm curious to see if Nemotron 4 will take them to a leading edge level instead of just briefly competitive.
(Nvidia released a Nemotron 3 and a Nemotron 4 like 3 years ago... this year's Nemotron 3 is entirely unrelated. Nvidia's naming schemes leave a little bit to be desired.)
There are a bunch of them they just don't get the attention because China has flooded the social media channels and is exceedingly good at drowning out the discourse with their benchmaxxed models.
AllenAI and IBM are two companies that release open weight models every couple of months. There are others if you look. OpenAI releases ML models on the regular (not LLMs).
The American open weight and open source AI/ML landscape is very healthy.
He said something to the effect of "I love open source and open models and we'll do open models when it makes sense and closed models when it makes sense" in a recent Q&A.
Given that his company has already released open models, I find it funny that, as you described it, his remark communicates absolutely nothing whatsoever. Not sure what the question was, but this was an artful non-answer.
Laguna is the most recent and capable one that comes to mind. In its size class it is not as "smart" in my experience as qwen 3.5 122 or DeepSeek v4 flash 0731 (all at q8), but it's also not terrible.
> But it makes sense that if you're a university researcher you are thinking about what's a model that will be open weight and developed over the long term and doesn't raise 'Chyna' concerns in Washington DC
Why phrase it "Chyna" when it's an actual legitimate concern?
review of AllenAI Olmo research team and commitment to OSS -- AI2 complete transparency including training data, code, intermediate checkpoints, and detailed logs for reproducibility and scientific rigor.
I'm interested to see where they want to land performance-wise (i.e. which point they choose on the scaling curve) and the niche they want to carve. They have a decent ways to scale beyond trinity large, in paticular on posttrain/RL before they are competitive with open-weights, especially internationally.
Deepseek is explicitly banned [1] at LLNL and I wouldn't be suprised if there's a blanket ban on all Chinese models. But nowadays models like tera/luna could fill this area of the pareto front, and LANL already runs openai models on their clusters [2]. Maybe it's in custom SFT/RL, for instrument control or sensitive topics? But you'll still have to compete with frontier models + a harness.
I would have also liked to see a carrot tied to their offer. It'll be hard to get teams to contribute RL gyms or curated text. But throw in a "we'll fund a postdoc/student to do that" and I think you'd have teams scrambling to apply.
I'd actually suggest a great starting point would be a local command reviewer LLM. Could ostensibly be a modern AV type thing. Particularly seeing this lately has driven the need home deeper to me: https://x.com/chrisbanes/status/2085341561609425230?s=20
An open weight tool call auto-reviewer, has all sorts of achievable scaling curve milestones.
That's interesting a locally hosted LLM would be banned. I'm assuming locally hosted is included. Do they think it's been trained to sabotage equipment?
I can imagine if the US were already doing that as a safeguard, they would assume their "adversaries" (to use Anthropic language) were doing the same as well, whether that were true or not, and therefore would not trust those models even if locally hosted.
Locally running LLM means nothing because "open weight" models are still inscrutable.
It's like bringing a dog home from the rescue and just hoping that it doesn't have the tendency to bite kids in the face. You just can't know. All you can do is try to add some new training telling it not to bite kids.
There's no mention of "LLM" nor "language". It does mention "foundation model" which includes LLMs but that also includes non-LLM architectures and non-text data. Many of the Genesis Initiative proposals answer "foundation model" call with non-LLM systems. All the FM's I know about currently in this sphere are non-LLMs. The "about gs1" page also does not mention "LLM" but does talk more about agentic harness and workflows. That description certainly sounds LLM'ish but describes a more rich system. I don't mean to suggest that LLMs will not be part of these "genesis open models" but as described, this will not result in a replacement for the "claude" or "codex" commands.
This is not the same thing, right? IIUC, the awards for what you linked have already been given out. There aren't awards for the linked initiative - I think that's just Argonne National Lab asking for volunteers to make their (ANL's) award money stretch further, right?
Do all these models have any significant architectural differences or training data sources? What are the factors going into the diversity of their performance?
It's funny: you can give the link to an LLM an ask it questions about TFA without reading it, but an actual human will go out of his/her way to tell you to RTFA :')
That sums up nothing, and parroting it some more doesn't make it more true, it just shows us the mindset and intellectual horizon of detractors. Brexit, Thiel's drooling over "balkanization" to Epstein, this constant stream of trash comments, all the same stupid cloth, it all gets the same "no".
The Australian Liberal Party (basically our version of conservative republicans) proposed the National Energy Guarantee policy in 2017, which inevitably failed due to the media and public’s relative literacy and tendency to turn policy names into acronyms.
It looks like they're taking applications for training data (due August 14th), so I think it's safe to say this is just an announcement of intent and a call for involvement vs. something that is readily available. Seems almost quaint in comparison to the strategy of sucking up every piece of data you can find anywhere on the Internet and feeding it to your LLM but I suspect their intent is to be more careful in what they train their model on.
I have no doubt companies like Microsoft, Amazon, and Google will rush to give them all the data they want in order to keep those government contracts flowing.
Instead of the department of energy striving to promote energy conservation and sustainable, non-polluting electricity generation, it is feeding the LLM craze. New department motto: "burn, baby, burn".
The DOE does a lot of "energy consuming" or less than environmentally friendly work, to include, historically, nuclear tests, and also has had ownership of some of the largest TOP500 supercomputers over the years.
Large compute projects such as an open language model aren't too far from their usual. You could easily argue the race to AGI is the closest thing to a modern Manhattan Project we've had in some time.
Whether that's a good allocation of resources is debatable, but from a national strategic perspective this makes sense, since private industry has pulled out of government contracts before in the LLM space (see Anthropic), this is just hedging their bets.
Mostly because it's generally a bad idea for government to try to compete with a brand new tech industry with hundreds of billions in private capital developing commercial models. If the American private industry does actually wash out vs Chinese open models there might be talent available for them to put money into, so maybe they are just preparing for that scenario in the meantime.
> Mostly because it's generally a bad idea for government to try to compete with a brand new tech industry with hundreds of billions in private capital developing commercial models.
I don't see why it's a bad idea if the models are as dangerous as this brand new tech industry claims they are.
The more dangerous this tech is, the better the idea looks. Can you explain?
Commoditizing AI models serves the interests of just about everybody except for a relative handful of people in San Francisco. The more decentralized control of the technology is, the more its benefits can be realized by businesses and individuals rather than becoming a black hole of monopolistic rent seeking.
I don’t quite understand what the argument is. Government does as a matter of fact operate on the frontier of new technology. (It’s how we got the web.) Why shouldn’t it, exactly?
Also, the US has been involved in AI research since the 1940s. So it’s not exactly a new thing.
we're about witness the realization that "here's a tech that can make us a whole bunch of money" is actually "here's tech that will establish the next hegemony." american companies may compete with chinese companies on the former. only the USG can compete with the PRC on the former.
The USG getting involved might actually harm US AI efforts. It's not just about money. Who would want to use Claude or ChatGPT if it were run by the US government? Yet these products are essential for gathering training data.
You're kidding yourself if you don't realize that the US gov can have access to any data they want in any of these US-based products. That's why they consider so important to "win" the development race for LLMs. It is a matter of continuing to access and control data that most of the world needs, as China has already closed the door to them.
The American attitude is generally to let private companies build up a new industry so it can create jobs and pay taxes. However, in the LLM race, the Chinese open weight playbook pretty much killed that. China has basically commoditized LLMs. Chinese models are good enough, so the race has come down to who can offer the cheapest tokens.
Chinese open weight models are great for this turn, but American private models generate orders of magnitude more cashflow. This cashflow = investment in training future models. It's unclear how Chinese open weight companies are going to compete in future rounds if they can't raise the same capital for training runs.
The American business model is exceedingly efficient at building large businesses from zero. I wouldn't dismiss it as just a jobs creation thing.
It’s unclear where American labs future capital will come from. They pretty much exhausted private options at that point and it’s not clear how successful an ipo would be at the current time
Obviously to me, I express things from my point of view.
Raising too much from debt is a bit dangerous if you plan to go public relatively soon and don’t have a good story for it (I don’t believe they have one). You can continue raising from VCs, but at some point the valuation and dilution starts to become a real issue, and will make your ipo even more difficult. Their options are pretty much limited to raising money from hyperscalers (with required compute spending, so more circular funding), which is what they are doing, but you cannot do that infinitely without having a good story to tell Microsoft/Google/Amazon investors. The market is more skeptical than it was a few months ago, I’m not convinced you can do that for years to come
I think as a counter point we continue to see investment and buildout. What do you mean the market is more skeptical? Of course the market doesn’t really have an opinion per se and aren’t all of these companies growing in valuation, revenues, and profits? At least the public ones.
Why are Anthropic's and OpenAI's annualized revenue about $50B each?
LLMs need massive amounts of compute to compete, so I wouldn't claim that the great (and leading, and likely to continue to lead) LLMs are commodities end-to-end, even if the non-executing-at-scale LLMs files and IP are commoditized. The execute, the compute, that is what breathes life into the model, which is otherwise weak or dead.
Dario doesn't have a problem with open weights, he just thinks that open weights should be tested prior to release so you aren't giving everyone a zero-day button or a "make a virus" button.
That's it. That's the whole stance. Most sane normal people agree with this stance, the techno-libertarian crowd find it egregiously offensive.
I think I found the item the grandparent commenter was presumably referring to - "Our position on open-weights models", posted by Amodei and dated July 27, 2026 - which includes:
> some people have even accused Anthropic of wanting to ban open-weights models as a means of protecting our business. Anyone who has read my past writing should know that I don’t regard such bans as a useful measure, but let me state it clearly so that there is no doubt:
*Anthropic has never advocated for a ban on open-weights models.*
However, as you said, it also says "All sufficiently capable models, open and closed, should go through mandatory safety testing."
It is. Depending on who Trump has fired or put in charge of a department it can be another shell that pumps out low quality crap. It might be the most valuable contribution on this thread.
I've had an extremely bad experience working with Department of Energy affiliated programmers in AI. By my invitation, they are part of our workflow and act as humans in the loop, but they have extremely bad habits of gaslighting and accusing people of schizophrenia rather than getting work done.
Here's an example[1] of the difference between what a U.S. Department of Energy employee adds to a ticket versus a private industry AI completing instructions as assigned.
This isn't some cherry-picked example, it's just what I happen to be dealing with right at this moment, happened just a couple of moments ago.
Can you explain the screenshot a little more? It just looks like you’re comparing the output of a chatbot and Claude Code about a log file. If it’s a metaphor, it went over my head, sorry!
Let’s distinguish a bit. There are political appointees (Trump’s government employees as you say) who are mostly upper management, and there are career civil servants (all the government scientists are under this category) who have a strong culture of apolitical dedication to the mission of their agency and to the American people and Constitution, regardless of who the current president is. And in the DOE labs in particular most (not all) of the scientists are actually employed as government contractors, but they have a similar non-partisan ethos.
That doesn’t necessarily mean there’s no need to be concerned with potential impact of policy and priority changes from the administration, but it does temper the threat model because the government employees you’re considering trusting have given oaths of office to protect and defend the Constitution.
> and there are career civil servants (all the government scientists are under this category)
US national lab scientists are not even civil servants. The labs themselves are run by a corporation under contract to the DOE and the scientists work for that corp. The managing corporation changes from time to time and the scientists transparently start working for whatever assumes the replacement. The land, the hardware, the buildings and any physical products are owned by the US gov't. To a very large extent, the intellectual output is set free to the world in the form of papers, presentations and to some small extent (eg compared to CERN) in the form of software.
Right, I did specifically say that most of the DOE scientists are contractors, but I concede the phrase “government scientist” is a bit ambiguous. I appreciate the extra detail you added. I think the distinction between political appointee and scientist/researcher stands.
As an added complication, some of the DOE labs do have civil servant scientists, for example National Energy Technology Lab and National Renewable Energy Lab are like 50/50 civil servants and contractors. And most of the funding arm of DOE are career civil servants. LANL, Sandia, Livermore, Argonne are all staffed by contractors
Ah but Mira Murati's new Inkling is Apache 2.0
But it makes sense that if you're a university researcher you are thinking about what's a model that will be open weight and developed over the long term and doesn't raise 'Chyna' concerns in Washington DC
[0] https://allenai.org/
There is no question this is true for the Chinese open LLMs. GPT-OSS had a small moment of interest, arguably it was a success as an open model for a while but it’s not relevant now. The early llamas were probably the most successful for their time.
Allenai / olmo was never relevant as far as I can tell. It’s not super helpful, especially as a sovereign government initiative to build an also-ran, they should be going for real relevance.
Meta might release something this year. X AI's Grok is still due to release a model, if Elon keeps to his word even if they only release a distilled version. Reflection AI has been quiet, but their access to compute is ramping up. Microsoft's MAI is considering releasing some open weight models which would be great to see!
Ilya's SSI is unlikely to release an open model since he's aiming for radical safety. That bet could pay off if the existing approach produces so much chaos within the next 10-20 years that some global ban is achieved and a super safe model is promoted as the compliant route.
We don't get many huge model releases though. I think it's harder and more expensive to safety align them. Even if you do, people will work around the safety and abuse the models. Plus it makes it even easier for Chinese companies to distill things that aren't as easy over filtered APIs.
There is a lot of internet propaganda to the effect that the US is simply unable to release open weight models or that China has so many more AI companies that the US is drowning in Chinese open weight models, but it's more like we're being careful and China doesn't care. If you host a model in China, it has to be censored and downloading any models requires you to provide your identity. Huggingface is banned there. When they release their open models in the west, they don't have to care whether the models are aligned in any way.
What? US laboratories are currently unable to contain their agents while doing security testing, and besides that, time and time again US labs seem to put short-term money above long-term safety.
Wasn't that literally why they tried to oust Altman from OpenAI, as he basically was 100% focused on profits and tried to cut down on safety across the board and lied to get his way?
> If you host a model in China, it has to be censored and downloading any models requires you to provide your identity.
I'm not disagreeing with that first part (obviously that's about inference hosting, not creating/training weights or hosting those weights), but the second part I'm not so sure about. AFAIK, ModelScope (which is the Huggingface in China) seems to allow downloads without verifying any identity and also hosts a bunch of abliterated weights.
Is it not illegal to "hack others" and "defeat protection/defensive systems" in the US already, including for both individuals and companies? Regardless if it was "by accident" or not?
Personally, I think we will one day come to see access to open weight models as an inalienable right to defense against tyranny, the way the second amendment is framed today. Just as encryption has become, which we similarly had to fight for in the 90s. I also understand that some regulation is sensible, but that doesn't automatically mean mandatory restricted or supervised access; any such restriction has to be extremely well-justified as essential for protecting the liberty of the people.
And as far as supervised access, whether or not identification is "handled by a third party" or "data is deleted after verification is complete" is immaterial; a citizen must not be required to trust their government. Any trust can and will be abused given enough time. Our systems must be trustless, and any expansion of government must be matched by an expansion in citizens' ability to check said government, in order to stand the test of time.
So supervised access seems completely off the table. And this can't just stop at access to models. Because linguistic analysis is a thing, and LLMs are scarily good at it (and existing non-AI solutions are still quite good given enough data), even the possibility that a government or other entity can save your messages means you've opened yourself up to deanonymization and surveillance. The chilling effect this has is undeniable, and the Supreme Court has made it clear that we cannot authorize government policy which creates chilling effects against essential liberties. Not to mention the possibilities that each category of users may be served subtly different models designed to influence them or constrain their agency/capability.
We're left with a situation where distributed access to capable open models is the only defense against a government or NGO which has access to billions of dollars of surveillance infrastructure and compute.
What specific examples of this do you have in mind? I can’t think of any liberal democratic countries converging on a combination of (a) single party rule, (b) nearly universal intrusion of state or party actors into private sector entities, (c) financial repression of private investments, and (d) the associated suppression of domestic consumption.
China is an authoritarian government and its policies have no more bearing on what people settle for than the currently socially unacceptable regime in the US.
In my opinion, if one lacks the motivation or resolve to fight for these rights, they should do so quietly and not attempt to patronize others who still stand by these rights as not being "realistic".
Besides that, it was also discovered that their suggested inference parameters were wrong and led to worse behavior. Eventually someone discovered these works best (so if you have the issue with looping right now, try these, helps a lot for me but not 100% still) and was also what the evals used apparently: temperature: 1.0, top_p: 1.0, top_k:20
Now we're waiting for RC3 which Poolside said will come at one point, and hopefully also brings down the size again NVFP4 weights + full context can load properly again even on "smaller" hardware.
But as a pure coding model, pretty good.
Deepseek v4 Flash 0731 is so much better if you can run it, though.
Grain of salt, I think I grabbed Laguna after they fixed the initial looping issues, didn't notice those, but there might've been other fixes since.
Yeah, this is my perspective too on Laguna S 2.1. Works amazingly for coding, pretty bad for pretty much anything else. I don't do a lot of advanced math, supposedly it's good for that too.
If you watch it think, which you can, unlike American closed models, you can steer it. You can provide a a massive rocket ship stratospheric boost to help it orient itself. You have no self correction, there is no multiplayer in American proprietary models.
Sure it's great having super powerful mystic oracles that have the "right" answers. But I love respect & revere the open thinking. No it's not automous. But it is brilliant. And it considers. A lot. Deeply. It chases. That to me is the most human of models, even as it falls far astray.
You should help it. You can. Unlike these vicious dark surfaces which yield and tell you nothing. I think this is the actual meta-core-super-point of "The session you cannot take with you" (link below). It's the session that does not care about you, will not interact with you, will not peer with you, that is a dead remote far off oracle to you. Fuck these "oracles". They are a plague against the human spirit. We should alloy humanity and AI to Augment Intellect (Engelbart). (To do less is species treason.) https://earendil.com/posts/session-portability/ https://news.ycombinator.com/item?id=49118781
Looks like on <https://arena.ai> agent arena (grouped by lab) Nvidia is 15/15 (much worse than Thinky and Mistral) and on text arena it's 18/27
On <https://openrouter.ai/models?order=most-popular> I definitely see usage though (probably mostly cause Nemotron 3 Ultra is free) the grouped order is DeepSeek, Tencent, Xiaomi, OpenAI, Z.ai, Nvidia
At this point, Nemotron 3 is really an 8 month old model series. That's when Nemotron 3 Nano was released, and the Nemotron 3 Super/Ultra models this year are obviously based on that recipe, mostly just bigger with a few tweaks here and there. Against today's models, no, not that interesting. Each of the Nemotron 3 models were briefly competitive when they launched, but never exceptional, and less competitive with each scale up. The fact that it took so long for Nemotron 3 Ultra to launch really hampered its competitiveness.
The Nemotron 3 series is extremely open about training recipes and training data, far more open than most open weight models, and that is valuable.
Before Nemotron 3, Nvidia had never released a single LLM that I would consider interesting at all, so Nemotron 3 was a big step up. The closest thing was Mistral NeMo, but a significant part of the credit there goes to the Mistral team, not Nvidia.
Given how much Nemotron 3 improved, I'm curious to see if Nemotron 4 will take them to a leading edge level instead of just briefly competitive.
(Nvidia released a Nemotron 3 and a Nemotron 4 like 3 years ago... this year's Nemotron 3 is entirely unrelated. Nvidia's naming schemes leave a little bit to be desired.)
AllenAI and IBM are two companies that release open weight models every couple of months. There are others if you look. OpenAI releases ML models on the regular (not LLMs).
The American open weight and open source AI/ML landscape is very healthy.
https://huggingface.co/unsloth/Laguna-S-2.1-GGUF
Why phrase it "Chyna" when it's an actual legitimate concern?
What's the concern with China?
Deepseek is explicitly banned [1] at LLNL and I wouldn't be suprised if there's a blanket ban on all Chinese models. But nowadays models like tera/luna could fill this area of the pareto front, and LANL already runs openai models on their clusters [2]. Maybe it's in custom SFT/RL, for instrument control or sensitive topics? But you'll still have to compete with frontier models + a harness.
I would have also liked to see a carrot tied to their offer. It'll be hard to get teams to contribute RL gyms or curated text. But throw in a "we'll fund a postdoc/student to do that" and I think you'd have teams scrambling to apply.
[1] https://hpc.llnl.gov/about-livermore-computing/ai-ml-lc/lc-l...
[2] https://www.energy.gov/nnsa/articles/nnsas-los-alamos-nation...
An open weight tool call auto-reviewer, has all sorts of achievable scaling curve milestones.
https://simonwillison.net/2026/Jun/10/if-claude-fable-stops-...
I can imagine if the US were already doing that as a safeguard, they would assume their "adversaries" (to use Anthropic language) were doing the same as well, whether that were true or not, and therefore would not trust those models even if locally hosted.
It's like bringing a dog home from the rescue and just hoping that it doesn't have the tendency to bite kids in the face. You just can't know. All you can do is try to add some new training telling it not to bite kids.
[0] https://commission.europa.eu/news-and-media/news/strengtheni...
Large compute projects such as an open language model aren't too far from their usual. You could easily argue the race to AGI is the closest thing to a modern Manhattan Project we've had in some time.
Whether that's a good allocation of resources is debatable, but from a national strategic perspective this makes sense, since private industry has pulled out of government contracts before in the LLM space (see Anthropic), this is just hedging their bets.
I don't see why it's a bad idea if the models are as dangerous as this brand new tech industry claims they are.
The more dangerous this tech is, the better the idea looks. Can you explain?
Also, the US has been involved in AI research since the 1940s. So it’s not exactly a new thing.
The government was involved in basic internet research. It didnt try to operate pets.com
The American business model is exceedingly efficient at building large businesses from zero. I wouldn't dismiss it as just a jobs creation thing.
It’s unclear to you, perhaps? But they’ll raise funds and/or debt as needed in the US capital markets as they have been doing.
> They pretty much exhausted private options at that point
I don’t think this is true. The evidence is that they keep raising funding for build.
> it’s not clear how successful an ipo would be at the current time
It’s always unclear, but also IPO success doesn’t necessarily translate into long term business success.
Raising too much from debt is a bit dangerous if you plan to go public relatively soon and don’t have a good story for it (I don’t believe they have one). You can continue raising from VCs, but at some point the valuation and dilution starts to become a real issue, and will make your ipo even more difficult. Their options are pretty much limited to raising money from hyperscalers (with required compute spending, so more circular funding), which is what they are doing, but you cannot do that infinitely without having a good story to tell Microsoft/Google/Amazon investors. The market is more skeptical than it was a few months ago, I’m not convinced you can do that for years to come
What do you mean by "basically"?
Why are Anthropic's and OpenAI's annualized revenue about $50B each?
LLMs need massive amounts of compute to compete, so I wouldn't claim that the great (and leading, and likely to continue to lead) LLMs are commodities end-to-end, even if the non-executing-at-scale LLMs files and IP are commoditized. The execute, the compute, that is what breathes life into the model, which is otherwise weak or dead.
I too can have $50B revenues by selling dollars for 50 cents each, and in the process I'll make a smaller loss than they do.
That's it. That's the whole stance. Most sane normal people agree with this stance, the techno-libertarian crowd find it egregiously offensive.
> some people have even accused Anthropic of wanting to ban open-weights models as a means of protecting our business. Anyone who has read my past writing should know that I don’t regard such bans as a useful measure, but let me state it clearly so that there is no doubt: *Anthropic has never advocated for a ban on open-weights models.*
However, as you said, it also says "All sufficiently capable models, open and closed, should go through mandatory safety testing."
https://www.anthropic.com/news/position-open-weights-models
Here's an example[1] of the difference between what a U.S. Department of Energy employee adds to a ticket versus a private industry AI completing instructions as assigned.
This isn't some cherry-picked example, it's just what I happen to be dealing with right at this moment, happened just a couple of moments ago.
[1] https://ibb.co/vCg2G1Dn
That doesn’t necessarily mean there’s no need to be concerned with potential impact of policy and priority changes from the administration, but it does temper the threat model because the government employees you’re considering trusting have given oaths of office to protect and defend the Constitution.
US national lab scientists are not even civil servants. The labs themselves are run by a corporation under contract to the DOE and the scientists work for that corp. The managing corporation changes from time to time and the scientists transparently start working for whatever assumes the replacement. The land, the hardware, the buildings and any physical products are owned by the US gov't. To a very large extent, the intellectual output is set free to the world in the form of papers, presentations and to some small extent (eg compared to CERN) in the form of software.
As an added complication, some of the DOE labs do have civil servant scientists, for example National Energy Technology Lab and National Renewable Energy Lab are like 50/50 civil servants and contractors. And most of the funding arm of DOE are career civil servants. LANL, Sandia, Livermore, Argonne are all staffed by contractors