GLM-4.6 Benchmark Details

GLM-4.6 currently shows benchmark results led by LiveCodeBench (16 / 118, score 84.50), AIME2025 (15 / 106, score 98.60), MMLU Pro (41 / 124, score 83).

Benchmark Results

GLM-4.6

Benchmark Results

综合评估

10 evaluations

Benchmark / mode

Score

Rank/total

MMLU Pro

Standard Mode

67 / 124

MMLU Pro

Thinking Enabled

41 / 124

GPQA Diamond

Standard Mode

132 / 175

GPQA Diamond

Thinking Enabled

66 / 175

GPQA Diamond

Thinking EnabledTools

82.90

58 / 175

LiveBench

Standard Mode

71.22

18 / 52

LiveBench

Thinking Enabled

71.22

18 / 52

HLE

Standard Mode

5.20

142 / 149

HLE

Thinking Enabled

17.20

108 / 149

HLE

Thinking EnabledTools

30.40

67 / 149

编程与软件工程

5 evaluations

Benchmark / mode

Score

Rank/total

LiveCodeBench

Standard Mode

77 / 118

LiveCodeBench

Thinking Enabled

82.80

22 / 118

LiveCodeBench

Thinking EnabledTools

84.50

16 / 118

SWE-bench Verified

Standard Mode

60 / 103

SWE-bench Verified

Thinking EnabledTools

60 / 103

数学推理

4 evaluations

Benchmark / mode

Score

Rank/total

AIME2025

Standard Mode

92 / 106

AIME2025

Thinking Enabled

98.60

15 / 106

AIME2025

Thinking EnabledTools

98.60

15 / 106

FrontierMath - Tier 4

Standard Mode

2.10

56 / 80

AI Agent - 工具使用

1 evaluations

Benchmark / mode

Score

Rank/total

Terminal-Bench

Thinking EnabledTools

40.50

12 / 35

Agent能力评测

2 evaluations

Benchmark / mode

Score

Rank/total

τ²-Bench

Thinking EnabledTools

75.90

20 / 40

τ²-Bench - Telecom

Thinking EnabledTools

27 / 35

指令跟随

1 evaluations

Benchmark / mode

Score

Rank/total

IF Bench

Thinking Enabled

27 / 27

AI Agent - 信息收集

1 evaluations

Benchmark / mode

Score

Rank/total

BrowseComp

Thinking EnabledTools

45.10

36 / 43

Compare with other models