OpenAI releases o3-mini: advanced reasoning at a fraction of the cost

OpenAI launched o3-mini on January 7 2026, following its announcement in December 2025. It is the company's smallest and most affordable reasoning model, designed to bring extended thinking capabilities to any developer building on the OpenAI API.

The standout feature is a three-tier effort system: low, medium, and high. At low effort the model responds quickly and cheaply; at high effort it reasons deeply and tackles hard problems like an expert. Despite being the "small" version, o3-mini in tests outperformed o1 — its more expensive predecessor — on both AIME (olympiad-level mathematics) and SWE-bench (real-world code bug resolution).

In practical terms, developers can now dial in exactly how much reasoning power they need per request, paying only for what they use. This dramatically lowers the barrier to building AI agents, coding assistants, and analytical tools without runaway costs. The release marks a shift: advanced reasoning stops being a premium luxury and becomes a standard building block in modern software development.