That’s my intern y’all 🥹

"We will not mince words - o3-deep-research and o4-mini-deep-research are probably the most powerful agentic LLMs in the world right now. This is thanks to the native support for MCP, Search and Code Interpreter..." x.com/Smol_AI/status…

“Truth in JSON only”

/v1/responses: the best way to shop 🙂‍↕️

To use a PDF with the OpenAI API, you can just pass its URL now—no upload needed! 📄🔗

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).

i spend so much time handcrafting the interface for my swift responses api library because i feel like interacting with ai ought to be beautiful it's a few lines of code that let you access literal magic, it should feel like casting a spell

I see some confusion on here about the size of this week. It will be very large. Big. Humungous even

GPT-5 is so cheap • GPT-5 $1.25/m input — $10/m output → Cheaper than Claude Sonnet • GPT-5-mini $0.25/m input — $2/m output → Cheaper than Flash • GPT-5-nano $0.05/m input — $0.40/m output → Cheaper than Flash-lite All with 90% cache discount

chat I think we are back

big if true (it's true)

🖐

🚀Really excited for GPT-5!! If you’re building with it, try our new prompt optimizer — it helps you tailor your prompts for the model

🚀

I spend my weekend debugging perf so you can keep vibecoding B2B saas 🫡

Fun Fact: Responses API was built specifically for a world where models would be doing complex tasks with many tool calls – glad to hear the folks at Augment Code are getting the most out of it!

This is why we do it. This is why we get up in the morning