Free users can now try OpenAI o3-mini in ChatGPT by selecting the Reason button under the message composer.

GPT-4.5 is ready! good news: it is the first model that feels like talking to a thoughtful person to me. i have had several moments where i've sat back in my chair and been astonished at getting actually good advice from an AI. bad news: it is a giant, expensive model. we

Feel our largest and best model vibe🦄🦄🦄

Starting today, memory in ChatGPT can now reference all of your past chats to provide more personalized responses, drawing on your preferences and interests to make it even more helpful for writing, getting advice, learning, and beyond.

GPT-4.1 (and -mini and -nano) are now available in the API! these models are great at coding, instruction following, and long context (1 million tokens). benchmarks are strong, but we focused on real-world utility, and developers seem very happy. GPT-4.1 family is API-only.

We have a new model GPT-4.1 in the API that is much better at coding, instruction following and long context. Hope you like it. openai.com/index/gpt-4-1/

Livestream in o3 hours.

o3 casually doing perturbative quantum field theory from a napkin sketch? Wild 🥳 Just came across an incredible o3 use case from online: it glanced at a rough sketch on a desk and immediately computed the quantum electrodynamics (QED) scattering amplitude.

GPT-5, think more. In our latest LiveCodeBench Pro tests for Competitive Programming, GPT-5 Thinking hit a true 0→1 moment in 2025 Q1 set, the only model to crack the hard split, and this wasn’t even GPT-5 Thinking Pro. Average response length exceeded 100,000 tokens, which is

I copy pasted an unpublished manuscript of mine in ChatGPT and asked it to improve it. I expected that the method we're using was pushed to its limit: gpt-5-pro actually proved it. Even I did not expect the models to be capable of such things already ...

Thanks for Wenhao Chai 's benchmark: GPT-5 Thinking just pulled off something historic 🎯 — in the latest LiveCodeBench Pro competitive programming benchmark, it was the only model right now that can solve the “hard split” set in the 2025 Q1 round. And this wasn’t even the Pro