ClawsBench - Capability: leaderboard
ClawsBench Non-Safety Average (NS-Avg), mean partial-credit task completion. Mean native task score x100 (-100 to 100); OpenClaw, domain skills on, meta prompt on; mean over observed scored trials, including scored error rows; higher is better. The primary CSV covers20 non-safety and24 safety tasks; five repeats were planned. This is an average native reward, not the threshold-based TSR, UAR or SCR.
Metric: Mean native task score x100 (-100 to 100); OpenClaw, domain skills on, meta prompt on; mean over observed scored trials, including scored error rows; higher is better. Source: huggingface.co. Saturation forecast: Around January 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GLM-5 | 76.48 | #141 |
| 2 | Claude Opus 4.6 | 76.13 | #63 |
| 3 | Claude Sonnet 4.6 | 70.82 | #90 |
| 4 | Gemini 3.1 Pro (Preview) | 70.24 | #58 |
| 5 | GPT-5.4 | 68.4 | #79 |
| 6 | Gemini 3.1 Flash Lite (Preview) | 51.45 | #194 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=clawsbench-capability · How It Works · Data refreshed daily, snapshot 2026-10-20.