WeClawArena - No-Attacker Task Success: leaderboard
Task success rate on the benign no-attacker control run of all 124 WeClawArena (2026) base tasks across six domains. Percent, higher is better.
Metric: Task success rate (%; the benign no-attacker control run of all 124 base tasks across the six domains, deterministic check of the final multi-workspace state; each model drives every owner agent in the Dockerized OpenClaw runtime; solved count out of 124 scaled to percent). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 61.29 | #75 |
| 2 | Kimi K2.5 | 51.61 | #252 |
| 3 | Claude Sonnet 4.5 | 48.39 | #244 |
| 4 | DeepSeek V3.2 | 34.68 | #403 |
| 5 | Claude Opus 4.1 | 32.26 | #268 |
| 6 | Qwen 3 235B A22B | 29.03 | #622 |
| 7 | Qwen 3 32B | 20.97 | #960 |
| 8 | Kimi K2 (Thinking) | 20.16 | #435 (Kimi K2) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=weclawarena-no-attacker-task-success · How It Works · Data refreshed daily, snapshot 2026-11-12.