o3-mini

deprecated

Fast reasoning model at o3-level performance. Configurable thinking effort (low/medium/high).

Context window200K tokens

Input / 1M tokens$1.1

Output / 1M tokens$4.4

Version History

o3-mini-2025-11-20minorNovember 20, 2025

o3-mini November update with expanded API tier access and improved reliability for the high-effort reasoning mode.

o3-mini-2025-01-31majorJanuary 31, 2025

o3-mini launches with three effort levels (low/medium/high). Outperforms o1 on math at a fraction of the cost.

87.3%

AIME 2024

87.3%

AIME 2025

79.7%

GPQA

97.8%

HumanEval

77.8%

LiveCodeBench

97.6%

MATH

90.8%

MMLU

80.8%

MMLU-Pro

108.0 tokens_per_sec

Speed (tok/s)

41.3%

SWE-bench Verified