An end-to-end real-world agent benchmark for OpenClaw-style workflows spanning tool use, planning, execution, and recovery across practical tasks.
As of March 2026, Claude Opus 4.6 leads the Claw-Eval leaderboard with 66.3% , followed by GPT-5.4 (66.3%) and Claude Sonnet 4.6 (66.3%).
Claude Opus 4.6
Anthropic
GPT-5.4
OpenAI
Claude Sonnet 4.6
Anthropic
According to BenchLM.ai, Claude Opus 4.6 leads the Claw-Eval benchmark with a score of 66.3%, followed by GPT-5.4 (66.3%) and Claude Sonnet 4.6 (66.3%). The top models are clustered within 0.0 points, suggesting this benchmark is nearing saturation for frontier models.
21 models have been evaluated on Claw-Eval. The benchmark falls in the Agentic category. This category carries a 22% weight in BenchLM.ai's overall scoring system. Claw-Eval is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
Year
2026
Tasks
Real-world agent workflows
Format
End-to-end agent evaluation
Difficulty
Broad real-world agentic execution
Claw-Eval is designed to test whether models can actually complete broad agent workflows instead of only local tool calls. It is useful for comparing agent reliability on realistic multi-step tasks with branching execution paths.
Claw-EvalVersion
Claw-Eval 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
An end-to-end real-world agent benchmark for OpenClaw-style workflows spanning tool use, planning, execution, and recovery across practical tasks.
Claude Opus 4.6 by Anthropic currently leads with a score of 66.3% on Claw-Eval.
21 AI models have been evaluated on Claw-Eval on BenchLM.
Get notified when new models drop, benchmark scores change, or the leaderboard shifts. One email per week.
Free. No spam. Unsubscribe anytime. We only store derived location metadata for consent routing.