Skip to main content

leaderboard.cn

Explore

Frontier Models
Historical Trends
Local Models
Compare Models
Ranking Studio

Library

Models
Pricing
Organizations
Benchmarks
Data sources

Loading page content

© 2026 Leaderboard.cn. Traceable AI model evaluations and rankings.

About Ranking methodology Changelog

v0.1.26.a85cb1563d03浙ICP备09076226号-7

Benchmark details loading

Back to benchmarks

Terminal Bench 2.1 (Best Harness)Coding

Official website

Claude Code 2.1.167, max_new_tokens=128k, 5 runs averaged

#ModelScore

1

GPT-5.6 SolOpenAI

88.8%

2

Kimi K3Moonshot AI

88.3%

3

Claude Opus 4.8Anthropic

84.6%

4

Claude Fable 5Anthropic

84.6%

5

83.4%

6

GLM-5.2Zhipu AI

82.7%

7

Gemini 3.1 ProGoogle

70.7%

8

GLM-5.1Zhipu AI

69.0%

8 models total