thunderkittens is now running on nvidia vera rubin! our kernels team got early access to nvl72 and spent the past few days digging through the new isa and bringing nvfp4 + fp8 gemms to life after reworking the kernels for rubin, we pushed them past 22 and 12 pflops respectively — competitive with cublas + cute dsl

It's exciting to see the rumors that Astra is a looped model. For those interested in the math on how to stably train them, I would check out our work Parcae. We also show that looping follows predictable scaling laws! arxiv.org/pdf/2604.12946

new: OpenAI & others quietly using loop transformers that don't show their 'thinking' when scaled up a leap forward on performance, but sparking concerns inside & outside OpenAI re: security as this takes off

Wow, the community is moving so fast. @reactorworld just developed an infinite video stream on Twitch using FastH3, and it looks so fun! It is open source!! Thank you @reactorworld for the love with OSS 🥰🥳 code and models here: github.com/reactor-team/infi… haoailab.com/blogs/fasth3-pr…

Excited to share that 0x Alpha will be on @togethercompute as early as tomorrow morning! together.ai/models/ox-alpha

.@Kimi_Moonshot benchmarked K3 endpoints across major inference providers. Together AI leads or ties for #1 on 3 of 4 benchmarks: OCRBench, MMMU Pro Vision, and DeepSWE. Open models like Kimi K3 have gotten genuinely good. Frontier-class good. Which makes serving them well essential. Proud of our Research team for holding the bar this high, and proud to see it verified by Moonshot. Run Kimi K3 on Together AI: togetherai.link/k3-x

(1/6) If a prompt can generate every layer of the stack, what are programming abstractions worth? GPU programming is at a weird inflection point. For decades we stacked layers of abstraction to ease our cognitive load. This year, we find ourselves constantly deleting them.

Some observations on Kimi: 1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also seemed very token hungry. It's not obvious to me that this model is actually that cheap to run. 2. I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks. To be clear, I *myself* might be fine with models presenting this level of marginal risk being open weight, but I am surprised that China is fine with it. I suspect the reason they are is 75% explained by strategic blindness/lack of AGI-pilledness (the CCP is very Yann Lecun-y in its views of AI). The other 25% or so is their lack of compute for customer inference (making China's open-weight strategy an unintended byproduct of US export controls) and the normal Chinese strategy of aggressive exports. For the companies, as opposed to the government, the decision to open source is partially ideological and partially because they are behind, and they know that very few people would pay for sub-frontier models from China. 3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex. 4. One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape, but I've never met an open-weight models advocate who doesn't ultimately concede this is where things end. You'd be surprised how many 'accelerationists' lobbied me, while I was in government, to support an eleven or twelve-figure federally funded data center so that startups could train models at a subsidy and then give them away for free. There was no other way for AI to progress, they said. Perhaps this is the logical end state of things. Nonetheless, I find myself surprised to see supposed accelerationists excited about such an outcome. I think many of them just don't know what they're doing. Many accelerationists do not view the creation and serving of frontier models as a legitimate business. 5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this. 6. It's probably true that open-weight models of this capability make the world a bit more dangerous, but not so much more that you'll really notice. At some point the models will be capable enough that you will notice. "A nonliving, invisible, dangerous, and infinitely self-replicating agent escaped from a Chinese lab," you say? Color me shocked.

next week at @aiDotEngineer, we are joining @togethercompute for a conversation on what goes into running agents at scale. @olive_jy_song, Research Lead, RL at MiniMax, and @realDanFu, VP of Kernels at Together AI, will walk through both sides of M3: the training decisions behind long-context reasoning and tool use, and the infrastructure work needed to serve agentic workloads in production. Wed July 1, 2:50 PM PDT Room 2016, Track 9 Catch you there!