Claude Opus 4 Benchmark Details

Claude Opus 4 currently shows benchmark results led by MATH-500 (4 / 46, score 98.20), MMLU-Pro (28 / 175, score 85), Aider-Polyglot (13 / 59, score 72).

Benchmark Results

Claude Opus 4

Benchmark Results

Knowledge Exams

5 evaluations

Benchmark / mode

Score

Rank/total

MMLU-Pro

Standard Mode

28 / 175

HLE

Standard Mode

10.70

389 / 568

HLE

Standard Mode

6.20

463 / 568

HLE

Thinking Enabled

12.30

369 / 568

HLE

Thinking Enabled

10.72

388 / 568

Abstract Generalization

2 evaluations

Benchmark / mode

Score

Rank/total

ARC-AGI-1

Standard Mode

35.67

146 / 176

ARC-AGI-2

Standard Mode

8.61

115 / 164

Scientific Reasoning

2 evaluations

Benchmark / mode

Score

Rank/total

GPQA Diamond

Standard Mode

79.60

242 / 461

GPQA Diamond

Thinking Enabled

79.60

242 / 461

Repository Engineering

1 evaluations

Benchmark / mode

Score

Rank/total

SWE-bench Verified

Standard Mode

72.50

53 / 116

Mathematics

11 evaluations

Benchmark / mode

Score

Rank/total

MATH-500

Standard Mode

98.20

4 / 46

AIME 2024

Standard Mode

35 / 61

AIME2025

Standard Mode

75.50

107 / 215

AIME2025

Thinking Enabled

73.30

113 / 215

FrontierMath

Standard Mode

4.50

39 / 60

FrontierMath

Thinking Enabled

4.10

41 / 60

FrontierMath - Tier 4

Standard Mode

72 / 80

FrontierMath - Tier 4

32K

4.20

40 / 80

FrontierMath - Tier 4

Thinking Enabled

4.20

40 / 80

IMO-ProofBench

Thinking Enabled

2.90

16 / 16

IMO-ProofBench Advanced

Thinking Enabled

2.90

24 / 24

Algorithmic Coding

2 evaluations

Benchmark / mode

Score

Rank/total

LiveCodeBench

Standard Mode

56.60

144 / 251

LiveCodeBench

Thinking Enabled

63.60

122 / 251

Writing

1 evaluations

Benchmark / mode

Score

Rank/total

Creative Writing

Standard Mode

1576.80

41 / 106

Visual Understanding

1 evaluations

Benchmark / mode

Score

Rank/total

MMMU

unknown

76.50

28 / 74

Commonsense

1 evaluations

Benchmark / mode

Score

Rank/total

SimpleBench

Thinking Enabled

58.80

39 / 96

Agentic Development

3 evaluations

Benchmark / mode

Score

Rank/total

Aider-Polyglot

Standard Mode

70.70

16 / 59

Aider-Polyglot

32K

13 / 59

Terminal Bench Hard

Thinking EnabledTools

31.10

105 / 244

Service Workflows

2 evaluations

Benchmark / mode

Score

Rank/total

τ²-Bench - Telecom

Thinking EnabledTools

73.40

129 / 264

τ²-Bench

Thinking EnabledTools

72.50

24 / 44

Instruction Following

2 evaluations

Benchmark / mode

Score

Rank/total

IF Bench

Standard Mode

43.30

191 / 282

IF Bench

Thinking Enabled

53.70

142 / 282

Capability Indices

1 evaluations

Benchmark / mode

Score

Rank/total

ECI

unknown

142.67

85 / 167

Compare with other models