OpenAI o3-mini (high) Benchmark Details

OpenAI o3-mini (high) currently shows benchmark results led by MATH (1 / 42, score 97.90), HumanEval (1 / 39, score 97.60), MATH-500 (9 / 44, score 97.90).

Benchmark Results

OpenAI o3-mini (high)

Benchmark Results

General Knowledge

3 evaluations

Benchmark / mode

Score

Rank/total

MMLU

86.90

26 / 66

GPQA Diamond

79.70

84 / 187

ARC-AGI

34.50

52 / 68

Coding and Software Engineer

3 evaluations

Benchmark / mode

Score

Rank/total

HumanEval

97.60

1 / 39

LiveCodeBench

69.50

55 / 123

SWE-bench Verified

49.30

95 / 112

Math and Reasoning

5 evaluations

Benchmark / mode

Score

Rank/total

MATH

97.90

1 / 42

MATH-500

97.90

9 / 44

AIME 2024

18 / 62

FrontierMath

23 / 60

FrontierMath - Tier 4

4.20

40 / 80

Common Sense

1 evaluations

Benchmark / mode

Score

Rank/total

SimpleQA

13.80

37 / 47

Compare with other models