o3
OpenAI🇺🇸 United States
OpenAI's most capable reasoning model. Leads on math, code, and science.
Context window200K tokens
Input / 1M tokens$10
Output / 1M tokens$40
Version History
o3-2026-01-10minor
o3 January 2026 update with expanded API availability across all tiers and improved performance on multi-step code debugging tasks.
o3-2025-04-16major
o3 opened to all API tiers. #1 on AIME 2025, SWE-bench, and Frontier Math.
Benchmark Scores
Full leaderboard →91.6%
AIME 2024
96.7%
AIME 2025
87.7%
GPQA
99.2%
HumanEval
83.1%
LiveCodeBench
97.9%
MATH
92.4%
MMLU
85.2%
MMLU-Pro
82.9%
MMMU
32.0 tokens_per_sec
Speed (tok/s)
71.7%
SWE-bench Verified