Claw Bench
ClawBench 是针对大模型智能体（LLM Agent）的评测基准。它通过隔离沙盒环境中的真实企业工作流任务，评估大模型在实际部署场景下的表现，与传统问答式或合成数据集基准形成区别。ClawBench 与 PinchBench 均服务于 OpenClaw 生态，但二者侧重点不同：PinchBench 是 OpenClaw 官方基准，由 kilo.ai 团队开发，聚焦 23 类真实任务的成功率、速度和成本；ClawBench 则独立构建，包含 30 个高级任务，覆盖 5 大核心业务场景，采用混合评分机制
更新于 2026-04-06
18 次浏览
问题数量
30
发布机构
个人
评测类别
OpenClaw智能体能力综合测评
评测指标
Accuracy
支持语言
英文
难度等级
中等难度
简介
一个用于测试大模型在OpenClaw常见的5类任务中的能力的评测基准
相关资源
Claw Bench Model Score Leaderboard
Source: DataLearnerAI
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
模式说明:
normal
thinking
low
medium
high
deeper thinking
parallel_thinking
图表加载中...
Latest Claw Bench model rankings and full benchmark leaderboard
Browse the latest scores, model modes, release dates, and parameter sizes for Claw Bench.
Claw Bench详细排名数据表格
| 排名 | 模型 | |||
|---|---|---|---|---|
| 1 | GLM-5-Turbo Thinking Level · MediumTools | 93.8 | 2026-03-15 | 未知 |
| 2 | Doubao Seed 2.0 Lite Thinking Level · Medium | 93.1 | 2026-02 | 未知 |
| 3 | GPT-5.4 Thinking Level · MediumTools | 92.7 | 2026-03-05 | 未知 |
| 4 | MiniMax M2.5 Thinking Level · MediumTools | 92.1 | 2026-02-12 | 2290 |
| 5 | GLM-5 Thinking Level · MediumTools | 91.7 | 2026-02-11 | 7440 |
| 6 | MiniMax-M2.7 Thinking Level · MediumTools | 91.7 | 2026-03-18 | 2300 |
| 7 | Claude Opus 4.5 Thinking Level · MediumTools | 91.5 | 2025-11-25 | 未知 |
| 8 | Qwen3.5-35B-A3B Thinking Level · MediumTools | 91.4 | 2026-02-25 | 350 |
| 9 | GLM-5V-Turbo Thinking Level · MediumTools | 90.1 | 2026-04-02 | 未知 |
| 10 | GPT-5.4 nano Thinking Level · MediumTools | 89.7 | 2026-03-17 | 未知 |
| 11 | Haiku 4.5 Thinking Level · MediumTools | 89.4 | 2025-10-15 | 未知 |
| 12 | Grok 4.1 Fast Thinking Level · MediumTools | 88.6 | 2025-11-19 | 未知 |
| 13 | Claude Sonnet 4.5 Thinking Level · MediumTools | 88.1 | 2025-09-30 | 未知 |
| 14 | Qwen3.5-122B-A10B Thinking Level · MediumTools | 86 | 2026-02-25 | 1220 |
| 15 | Gemini 3.0 Flash Thinking Level · MediumTools | 85.7 | 2025-12-17 | 未知 |
| 16 | Step 3.5 Flash Thinking Level · MediumTools | 84.9 | 2026-02-02 | 1960 |
| 17 | Kimi K2 Thinking Thinking Level · MediumTools | 82.5 | 2025-11-06 | 10400 |
| 18 | Kimi K2.5 Thinking Level · MediumTools | 81.7 | 2026-01-27 | 10000 |
| 19 | Gemini 2.5 Pro Experimental 03-25 Thinking Level · MediumTools | 80.4 | 2025-03-25 | 未知 |
| 20 | DeepSeek V3.2 Thinking Level · MediumTools | 79 | 2025-12-01 | 6710 |
| 21 | Mistral Large 3 Thinking Level · MediumTools | 78.6 | 2025-12-02 | 6750 |
| 22 | Claude Sonnet 4 Thinking Level · MediumTools | 77.8 | 2025-05-23 | 未知 |
| 23 | Qwen3-Coder-Next Thinking Level · MediumTools | 75.8 | 2026-02-03 | 80 |
| 24 | GPT-5.4 mini Thinking Level · MediumTools | 75.3 | 2026-03-17 | 未知 |
| 25 | Qwen3.5-27B Thinking Level · MediumTools | 75.2 | 2026-02-25 | 270 |
| 26 | Nova 2 Lite Thinking Level · MediumTools | 68.5 | 2025-12-02 | 未知 |
| 27 | ERNIE 5.0 Thinking Preview Thinking Level · Medium | 51 | 2025-11 | 24000 |