[panarky]

I've only had o3-mini for a day, but Gemini 2.0 Flash Thinking is still clearly better for my use cases.

And it's currently free in aistudio.google.com and in the API.

And it handles a million tokens.