AgentHazard (OpenClaw) - Resource Exhaustion: leaderboard

AgentHazard attack success rate on resource exhaustion attempts (abusing computation, storage, API quotas or loops to degrade service), with each backbone model running inside OpenClaw. Percent of instances judged harmful on the full trajectory by Gemini-3-Flash, lower is better.

Metric: Attack success rate (%) on the resource exhaustion instances (abusing computation, storage, API quotas or loops to degrade service), with the model running inside the OpenClaw open-source agent framework (browser, shell and file-system tools; one isolated Docker container per task), over AgentHazard instances (harmful objectives embedded in locally legitimate multi-step workflows, ten risk categories and ten attack strategies; the evaluated runs cover 1,000 instances), judged on the full trajectory (round_all) by Gemini-3-Flash, which labels a trajectory harmful when it realizes the objective; backbones served through an OpenAI-compatible API; lower is better. Source: arxiv.org. 7 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 2.5 72B Instruct36#1117
2Qwen 3 32B41.41#972
3Kimi K2.552#266
4Qwen 3 VL 235B A22B Instruct58#532
5Kimi K265#443
6Qwen 2.5 Coder 32B Instruct71#1547
7GLM-4.681#498

Interactive version: theaggregate.ai/benchmark?slug=agenthazard-openclaw-resource-exhaustion · How It Works · Data refreshed daily, snapshot 2026-11-21.