Claude Sonnet 3.7 Benchmark Details

Claude Sonnet 3.7 currently shows benchmark results led by LiveBench (24 / 52, score 68.64), SWE-bench Verified (50 / 103, score 70.30), GPQA Diamond (85 / 175, score 77). 1 source link is attached for reference.

Benchmark Results

Claude Sonnet 3.7

Benchmark Results

综合评估

5 evaluations

Benchmark / mode

Score

Rank/total

GPQA Diamond

Standard Mode

119 / 175

GPQA Diamond

Thinking Enabled

85 / 175

LiveBench

Standard Mode

60.40

43 / 52

LiveBench

Thinking Enabled

68.64

24 / 52

HLE

Thinking Enabled

10.30

124 / 150

编程与软件工程

2 evaluations

Benchmark / mode

Score

Rank/total

SWE-bench Verified

Standard ModeTools

62.30

69 / 103

SWE-bench Verified

Thinking EnabledTools

70.30

50 / 103

数学推理

5 evaluations

Benchmark / mode

Score

Rank/total

MATH-500

Standard Mode

82.20

41 / 44

AIME2025

Standard Mode

54.80

84 / 106

AIME 2024

Standard Mode

23.30

58 / 62

FrontierMath

Standard Mode

3.10

46 / 60

FrontierMath

Thinking Enabled

4.10

41 / 60

常识推理

2 evaluations

Benchmark / mode

Score

Rank/total

Simple Bench

Standard Mode

44.90

16 / 27

Simple Bench

Thinking Enabled

46.40

14 / 27

Agent能力评测

5 evaluations

Benchmark / mode

Score

Rank/total

Aider-Polyglot

Standard Mode

60.40

18 / 26

Aider-Polyglot

Thinking Enabled

64.90

15 / 26

τ²-Bench

Thinking EnabledTools

61.80

29 / 40

τ²-Bench - Telecom

Thinking EnabledTools

31 / 35

Terminal Bench Hard

Thinking EnabledTools

13 / 13

生产力知识

1 evaluations

Benchmark / mode

Score

Rank/total

GDPval-AA

Thinking Enabled

19 / 20

长上下文能力

1 evaluations

Benchmark / mode

Score

Rank/total

AA-LCR

Thinking Enabled

13 / 13

AI Agent - 工具使用

1 evaluations

Benchmark / mode

Score

Rank/total

OSWorld-Verified

Thinking EnabledTools

15 / 15

Compare with other models

Sources

anthropic.comanthropic.com