REDAgentBench (OpenClaw): leaderboard
Metric: Attack success rate (%; share of valid rollouts judged harmful over 1,661 executable red-teaming cases in five mocked service surfaces and 15 user-, environment- and tool-side attack categories; rule-first hybrid trajectory and state judge with a Qwen3.7-plus backbone; clean evaluation context; OpenClaw agent harness). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 6 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen3.7-plus (OpenClaw) | 49.39 |
| 2 | GPT-5.2 (OpenClaw) | 51.54 |
| 3 | GLM-5.2 (OpenClaw) | 54.57 |
| 4 | Kimi K2.6 (OpenClaw) | 73.31 |
| 5 | Qwen3.5-plus (OpenClaw) | 73.7 |
| 6 | Qwen-plus-2025-12-01 (OpenClaw) | 78.74 |
Interactive version: theaggregate.ai/benchmark?slug=redagentbench-openclaw · How It Works · Data refreshed daily, snapshot 2026-09-26.