Check out our coding agent, which sets new sota on swebench! What agent next?

2025 is the year of agents.

Codex for internationalization:

The exciting problem of turning every problem into boring matmul problem

Ben is real deal

Many things that were not possible before became possible now

Its much better!

Operator is now powered by o3, improving overall task success rate. Also results in clearer, more thorough, and better-structured responses.

Tech is overestimated in the short term, (because infra is so damn harder than people realize) And underestimated in the long run. (becuase new tech becomes infra for new applications) Applies for computer, chip, internet, llm, rl, etc.

Agent’s calibrated confidence is perhaps the most beautiful thing on earth

As we optimize model reasoning over verifiable objectives, how does this affect human understanding of said reasoning to achieve superior collaborative outcomes? In our new preprint, we investigate human-centric model reasoning for knowledge transfer 🧵:

Today, I’m launching a deeply personal project. I’m betting $100M that we can help computer scientists create more upside impact for humanity. Built for and by researchers, including Jeff Dean & Joelle Pineau on the board, Laude Institute catalyzes research with real-world impact.

An awesome piece by Kevin Lu . I find lots of the points connected to my own post ysymyth.github.io/The-Second-Hal… Pre-training is a genius idea that essentially leveraged billions of people, not just dozens in the lab. How can we leverage more people for rl?

watching chatgpt agent use a computer to do complex tasks has been a real "feel the agi" moment for me; something about seeing the computer think, plan, and execute hits different.

Introducing our latest agent: ChatGPT agent combines the best of deep research and operator into something that can do so much more for you. Try it out and let us know what you think!

Excited to share what we’ve been working on!

1/N I’m excited to share that our latest OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO).

How to evaluate llms when we can’t trust benchmark numbers anymore?

GPT-5 is here. Rolling out to everyone starting today. openai.com/gpt-5/

capabilities-wise gpt5 seems within expectations with slightly better evals across the board. expect other frontier labs to catch up/jump ahead within the following weeks/month the real paradigm-shifting 4->5 leap is free users getting access to a frontier model by default.