OpenAI o1 Graduates to General Availability

OpenAI's reasoning model o1, which had been in a limited preview since late 2024, is now generally available to all API users. The upgrade is substantial. When it launched, o1 was stripped-down: no vision, no tools, no system prompt, just raw reasoning. That was frustrating for developers who wanted to use it in real apps. Now it has all of those: you can send images and ask it to reason about them, you can give it tools to call, and you can set a system prompt to define its persona or context. The context window jumped to 200K tokens — big enough for very long documents or codebases. The price drop is significant: 50% cheaper than the preview, which already wasn't cheap, making serious production use much more viable. o1 is still the go-to choice when you need careful multi-step reasoning — complex math, scientific analysis, tricky code review — and now it's actually practical to build with.