Claude Sonnet 4 Benchmark Details

Claude Sonnet 4 currently shows benchmark results led by SWE-bench Verified (14 / 116, score 80.20), MMLU-Pro (40 / 175, score 84), Terminal-Bench (10 / 35, score 41.30). 1 source link is attached for reference.

Benchmark Results

Claude Sonnet 4

Benchmark Results

Knowledge Exams

5 evaluations

Benchmark / mode

Score

Rank/total

MMLU-Pro

Thinking Enabled

40 / 175

HLE

Standard Mode

5.52

473 / 568

HLE

Standard Mode

4.30

514 / 568

HLE

Thinking Enabled

10.70

389 / 568

HLE

Thinking Enabled

9.60

402 / 568

Abstract Generalization

4 evaluations

Benchmark / mode

Score

Rank/total

ARC-AGI-1

Standard Mode

23.80

155 / 176

ARC-AGI-1

Thinking Enabled

140 / 176

ARC-AGI-2

Standard Mode

1.30

151 / 164

ARC-AGI-2

Thinking Enabled

5.93

123 / 164

Scientific Reasoning

5 evaluations

Benchmark / mode

Score

Rank/total

GPQA Diamond

Standard Mode

338 / 461

GPQA Diamond

Thinking Enabled

75.40

277 / 461

GPQA Diamond

DeepTools

83.80

182 / 461

CritPt

Standard Mode

1.10

151 / 204

CritPt

Thinking Enabled

0.30

186 / 204

Repository Engineering

3 evaluations

Benchmark / mode

Score

Rank/total

SWE-bench Verified

Thinking EnabledTools

72.70

52 / 116

SWE-bench Verified

Thinking Level · HighTools

80.20

14 / 116

SWE-Bench Pro - Public

Thinking Enabled

42.70

55 / 62

Mathematics

8 evaluations

Benchmark / mode

Score

Rank/total

AIME2025

Standard Mode

171 / 215

AIME2025

Thinking Enabled

70.50

119 / 215

AIME2025

DeepTools

77 / 215

AIME 2024

Standard Mode

43.40

50 / 61

IMO-ProofBench

Thinking Enabled

27.10

8 / 16

IMO-ProofBench Advanced

Thinking Enabled

4.80

19 / 24

FrontierMath

Standard Mode

4.10

41 / 60

FrontierMath - Tier 4

Standard Mode

72 / 80

Algorithmic Coding

7 evaluations

Benchmark / mode

Score

Rank/total

CodeClash

Standard ModeTools

1223

4 / 8

LiveCodeBench

Standard Mode

48.50

171 / 251

LiveCodeBench

Thinking Enabled

106 / 251

IOI 2024 (Vals v1)

Standard ModeTools

9.70

5 / 10

IOI 2024 (Vals v1)

Thinking EnabledTools

5.20

7 / 10

IOI 2025 (Vals v1)

Standard ModeTools

3.30

6 / 9

IOI 2025 (Vals v1)

Thinking EnabledTools

5 / 9

Writing

1 evaluations

Benchmark / mode

Score

Rank/total

Creative Writing

Standard Mode

1480.30

53 / 106

Agentic Development

8 evaluations

Benchmark / mode

Score

Rank/total

Aider-Polyglot

Standard Mode

56.40

26 / 59

Aider-Polyglot

32K

61.30

20 / 59

Terminal-Bench

Standard ModeTools

26 / 35

Terminal-Bench

Thinking EnabledTools

35.50

18 / 35

Terminal-Bench

DeepTools

41.30

10 / 35

Terminal-Bench 2.1

Thinking EnabledTools

36.30

157 / 199

Terminal Bench Hard

Standard ModeTools

27.30

120 / 244

Terminal Bench Hard

Thinking EnabledTools

31.10

105 / 244

Visual Understanding

4 evaluations

Benchmark / mode

Score

Rank/total

MMMU

Standard Mode

76.50

28 / 74

MMMU

unknown

74.40

34 / 74

MMMU-Pro

Standard Mode

62.40

165 / 229

MMMU-Pro

Thinking Enabled

61.80

170 / 229

Commonsense

1 evaluations

Benchmark / mode

Score

Rank/total

SimpleBench

Thinking Enabled

45.50

63 / 96

Service Workflows

6 evaluations

Benchmark / mode

Score

Rank/total

τ²-Bench - Telecom

Standard ModeTools

52.30

169 / 264

τ²-Bench - Telecom

Thinking EnabledTools

150 / 264

τ²-Bench

Standard ModeTools

35 / 44

SAGE

Standard ModeTools

49 / 64

SAGE

Thinking EnabledTools

52 / 64

τ³-Banking

Thinking EnabledTools

16.70

102 / 167

Instruction Following

3 evaluations

Benchmark / mode

Score

Rank/total

IF Bench

Standard Mode

45.40

180 / 282

IF Bench

Thinking Enabled

54.70

137 / 282

IF Bench

Thinking EnabledTools

134 / 282

Cross-capability Suites

2 evaluations

Benchmark / mode

Score

Rank/total

LiveBench

Standard Mode

50.98

91 / 117

LiveBench

64K

61.27

67 / 117

Cross-industry Work

1 evaluations

Benchmark / mode

Score

Rank/total

GDPval-AA

Thinking Enabled

13 / 15

Long Reasoning

2 evaluations

Benchmark / mode

Score

Rank/total

AA-LCR

Standard Mode

154 / 174

AA-LCR

Thinking Enabled

70.30

104 / 174

Desktop Workflows

1 evaluations

Benchmark / mode

Score

Rank/total

OSWorld-Verified

Thinking EnabledTools

42.20

26 / 28

Tool Orchestration

3 evaluations

Benchmark / mode

Score

Rank/total

Pinch Bench

Thinking EnabledTools

80.50

23 / 38

Claw Bench

Thinking EnabledTools

77.80

23 / 29

PinchBench v2

unknown

48.78

39 / 45

Clinical Workflows

4 evaluations

Benchmark / mode

Score

Rank/total

MedScribe

Standard ModeTools

72.41

57 / 66

MedScribe

Thinking EnabledTools

69.35

61 / 66

MedCode

Standard ModeTools

33.94

54 / 64

MedCode

Thinking EnabledTools

34.96

52 / 64

Capability Indices

1 evaluations

Benchmark / mode

Score

Rank/total

ECI

unknown

141.69

89 / 167

Compare with other models

Sources

artificialanalysis.aiartificialanalysis.ai