January 31, 2025
OpenAI o3-mini released
OpenAIAn AI research company — created GPT, ChatGPT, DALL-E, and the o-series reasoning models. released o3-mini on January 31, 2025 — a cost-efficient reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. model with adjustable effort levels, available to all ChatGPTOpenAI's conversational chat product — brought large language models to hundreds of millions of users. users including free tier.
What it was for
o3-mini delivered o3-class STEM reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. at o1-mini pricing — tripling Plus rate limits and answering DeepSeek R1's open-weight challenge weeks after R1's launch. It shipped with function calling, structured outputs, and web search before full o3 arrived in April.
Companies
- OpenAI
Why it's here
o3-mini brought frontier reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. to free ChatGPTOpenAI's conversational chat product — brought large language models to hundreds of millions of users. users at mini-model cost.
Why it mattered
Adjustable reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. effort made test-time compute tunable per task.
What it solved
Developers wanted o3-class math and coding without full o3 latency and price.
Media
- ImageOpenAI
OpenAI, Public domain, via Wikimedia Commons
Related
- OpenAI o3 reasoning models announcedDecember 20, 2024
- OpenAI o3 and o4-mini releasedApril 16, 2025
- DeepSeek R1 releasedJanuary 20, 2025