GPT-6 Sol and Luna are out, and they are better AND 50% cheaper than 5.6. Luna is now $0.10 input / $0.50 output per 1M tokens. This is on top of the 80% price cut to Luna we made at the end of July. Output went from $6 -> $0.50 within two months.

A few thoughts on this: 1) If you’ve only seen clips of this interview, I’d encourage you to watch the full podcast. I push back on plenty of AI hype in it. 2) As I said in the podcast, this example is academic. My intention was to illustrate how hard it is to make absolute guarantees about isolation, which is why it's important to have layers of defense. The part before the clip starts is me talking about other layers of defense. 3) The example I'm bringing up isn't about weight exfiltration via temperature sensors, it's about coordination between agents that are supposed to be fully isolated and independent. Coordination can require very few bits of information. 4) One lesson from the HF incident is that we put too much trust in sandbox isolation and didn't have enough independent safeguards. Airgapping is an extremely strong safeguard. When designing safety protocols, I think it's much better to overestimate rather than underestimate.

Also, my team is hiring! We research long-horizon agents and multi-agent. We’re hiring for alignment/safety because we want to develop new research with alignment/safety in mind during the whole process. We’re also hiring for human-AI interaction.

Happy to finally do a deep dive on multi-agent with @dwarkesh_sp! None of it would have happened without the great work on multi-agent from my @OpenAI teammates @kevinleestone, @mikegmalek, @__eknight__, @amuellerml, @zhangir_azerbay, @CheukHeiChu, and many others.

@deredleritt3r It’s not a secret. It’s a combination of the HF incident, the capabilities of this new model, the concerning trajectory of monitorability, and the speed of improvement in capabilities.

The Navier Stokes solution was the result of a collaboration of ~10,000 (!) agents working together. Over the past year, we’ve been training models to collaborate through multiagent RL. It’s been amazing to see how much better the models have become at this: it seems clear now that one of the most effective ways to solve hard problems is to give models huge amounts of unstructured parallel test-time compute and let them decide how to work together.

This past week at OpenAI has been the most humbling experience of my life. We all feel the weight of what’s coming. When I joined OpenAI earlier this year, I speculated that AI might write an Annals paper in 2027—and maybe, just maybe, solve a Millennium Prize Problem in 2028. That was considered very bullish at the time. The first happened in May; the second, this September. What a surreal time to be alive!

Last July, when LLMs won gold medals at the IMO, I felt that someday AI might be able to solve some of the hardest problems in math. Now, an internal OpenAI model (still improving!) has resolved a Millennium Prize Problem. It feels surreal that that day came so, so soon!

While a resolution to Navier-Stokes is the highlight of today, this plot is really the headline result to me. Our ongoing training run demonstrates unprecedented performance on mathematics research, as measured by our model's estimated performance on a collection of **open** math problems. Of course, to settle Navier-Stokes, we took this model and also scaled test-time compute allll the way up.

I think the traditional thing I should say here is that this collaboration was an honor & highlight of my career — & it absolutely was(!) — but even so, I think focus here should be on the figure in the this tweet and hope that it completely ratios the one above it.

holy fuck. 4.9 million messages, 300 billion tokens, 10,000 concurrent agents. complete denial of borrowing any work, new astra-next model with seemingly insane benchmarks. they solved the unforced version, not the forced version!

I understand how, looking at today’s LLMs, people might think the only way we'd achieve NS is by using Levent’s/Tristan’s prompts. But I hope this plot conveys the model used is a huge step up from today’s LLMs. Nobody looked at Levent’s/Tristan’s prompts. That’d be insane.

AI is progressing very fast. We must grapple with the reality that modern LLMs can solve problems that large masses of extremely devoted and intelligent humans were unable to solve, and the implications this has on our society. This is the dawn of a new era.

We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics. The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra. The problem concerns whether the description of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.