o3-mini

A smaller variant of O3, expected to offer enhanced multimodal capabilities, improved reasoning, and more efficient resource utilization compared to previous models while maintaining strong performance on core tasks.

COLLIE

98.7%

i
MATH

97.9%

i
IFEval

93.9%

i
MGSM

92.0%

i
AIME 2024

87.3%

i
MMLU

86.9%

i
LiveBench

84.6%

i
Multilingual MMLU

80.7%

i
Multi-IF

79.5%

i
GPQA

77.2%

i
Aider-Polyglot

66.7%

i
Aider-Polyglot Edit

60.4%

i
Graphwalks parents <128k

58.3%

i
TAU-bench Retail

57.6%

i
Graphwalks BFS <128k

51.0%

i
MultiChallenge (o3-mini grader)

50.2%

i
Internal API instruction following (hard)

50.0%

i
SWE-Bench Verified

49.3%

i
Multi-Challenge

39.9%

i
TAU-bench Airline

32.4%

i
OpenAI-MRCR: 2 needle 128k

18.7%

i
SWE-Lancer

18.0%

i
ComplexFuncBench

17.6%

i
SimpleQA

15.0%

i
FrontierMath

9.2%

i
SWE-Lancer (IC-Diamond subset)

7.4%

i

Pricing, uptime, and speed via OpenRouter — updated Jul 17, 2026, 04:19 AM.

Provider	Status	Input	Output	Limits	Uptime	Speed	Notes
OpenAI	available	$1.10/Mtok cache $0.55/Mtok	$4.40/Mtok	200K tokens context 100K tokens max output	—	2,026 ms p50 TTFT 214 tok/s p50	$0.01/web search