OpenClawProBench: leaderboard
OpenClawProBench open-profile reported average task score for each provider, platform and runtime deployment. Native means and source execution statuses are retained; this metric is distinct from the separately defined FinalScore composite and closed holdout protocols.
Metric: Reported Average Score (%). Source: suyoumo.github.io. Saturation forecast: Rough model projection: around 2026. 71 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K3 (ClawProBench Run: provider=moonshot; platform=kimi; runtime=v2026.6.11) | 81.61 | |
| 2 | GLM-5.2 (ClawProBench Run: provider=glm; platform=glm; runtime=3.2.6) | 81.26 | |
| 3 | agnes-2.0-flash (ClawProBench Run: provider=agnes; platform=agnes; runtime=3.2.6) | 77.01 | |
| 4 | intern-s2-preview (ClawProBench Run: provider=ShangHai%20AILab; platform=intern; runtime=3.2.6) | 76.71 | |
| 5 | MiniMax-M3 (ClawProBench Run: provider=minimax; platform=minimax; runtime=3.2.6) | 75.15 | |
| 6 | Sensenova 6.7 Flash Lite (ClawProBench Run: provider=sensetime; platform=sensenova; runtime=3.2.6) | 73.72 | |
| 7 | step-3.7-flash (ClawProBench Run: provider=stepfun; platform=stepfun; runtime=3.2.6) | 72.01 | |
| 8 | ERNIE 5.1 (ClawProBench Run: provider=baidu; platform=baiduqianfan; runtime=3.2.6) | 70.74 | |
| 9 | qwen3.5-397b-a17b (ClawProBench Run: provider=qwen; platform=bailiantokenplan; runtime=3.2.6) | 70.39 | |
| 10 | qwen3.5-plus (ClawProBench Run: provider=qwen; platform=bailiancodingplan; runtime=3.2.6) | 70.1 |
Interactive version: theaggregate.ai/benchmark?slug=openclawprobench · How It Works · Data refreshed daily, snapshot 2026-10-20.