Summary updated: 2026-08-28 · Curation only — we do not reproduce full original content.
Key takeaways
- API changelogs are the real news. OpenAI’s blog often announces rate-limit changes, new model IDs (e.g.,
gpt-4o), and deprecation timelines before the docs catch up. If you operate in production, the blog is the canary. - Function calling and tool use get their own series. Step-by-step posts on structured outputs, parallel function calling, and the Assistants API are where OpenAI concentrates its “how to build” content. These are more useful than generic release notes.
- Safety and moderation are documented, not optional. OpenAI publishes moderation API guides, red-teaming results, and usage policies. The blog frames these as product features; the docs treat them as integration requirements.
- Plugins are legacy; Assistants and Responses API are current. The blog occasionally references legacy plugins for backward compatibility, but new development should focus on the Responses API and tool-use patterns.
Official source
Developer Blog: https://developers.openai.com/blog
API Reference: https://platform.openai.com/docs
We summarize and curate only. All original content, code samples, and screenshots belong to OpenAI. Confirm current prices, model IDs, and terms on their official docs.