Michelle Pokrass
@michpokrass
shipping at openai
gpt-4.1 landing in chatgpt today!! we were initially planning on keeping this model api only but you all wanted it in chatgpt :) happy coding!
By popular request, GPT-4.1 will be available directly in ChatGPT starting today. GPT-4.1 is a specialized model that excels at coding tasks & instruction following. Because it’s faster, it’s a great alternative to OpenAI o3 & o4-mini for everyday coding needs.
love watching chatgpt agent do its thing, feels like a glimpse into the future. @isafulf and team cooked!
ChatGPT can now do work for you using its own computer. Introducing ChatGPT agent—a unified agentic system combining Operator’s action-taking remote browser, deep research’s web synthesis, and ChatGPT’s conversational strengths.
sooo many developers have asked for this -- excited to now support more features from jsonschema, like ranges for values, regexes, and fixing errors on parallel function calling
We've made some improvements to Structured Outputs: 🎣 Parallel function calling now works with strict mode—ensuring calls reliably adhere to schema ⚙️ Many more keywords are now supported, letting you specify: - Output string lengths and formats via regex or formats like email…
had a blast chatting with jacob about gpt-4.1!! learn a little more about how we built it and the latest on how to make agents work well in prod
Super fun to get one of @OpenAI 's GPT 4.1 Leads @michpokrass's thoughts on RFT (launched yesterday) and more on Unsupervised Learning. Some of my favorite parts: ▪️How OpenAI built GPT-4.1 by focusing on developer feedback over benchmarks ▪️Where agents work and what needs to…
structured outputs inspo! super cool use case from consensus which is uniquely enabled by structured outputs.
How @ConsensusNLP uses GPT-4.5 for nuanced scientific and medical written analysis, and Structured Outputs to visualize levels of agreement across research papers:
eager for your feedback for o3-mini in the api! we think it's a good model. let us know what can be better, and we'll fix it in the next rev. one thing top of mind already -- this model is a little harder to steer with function calling and can be a bit hesitant to call tools
the mandate of heaven is given to those who yearn for it, not those who believe it to be theirs