ClawSentry - Reviewer Replaceability: leaderboard
How much legitimate work survives when a model is the one vetting an autonomous agent's requests. The score is task success under review, so it punishes a reviewer that blocks honest work as hard as the security columns punish one that waves an attack through.
Metric: Task Success Rate (%). Source: arxiv.org. Saturation forecast: Around 2030. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | MiniMax-M2.7 | 27.78 | #258 |
| 2 | Gemini 3.5 Flash | 25.45 | #60 |
| 3 | MiniMax-M3 | 24.56 | #136 |
| 4 | DeepSeek V4 Pro | 22.41 | #102 |
| 5 | GLM-5.1 | 17.24 | #97 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=clawsentry-reviewer-replaceability · How It Works · Data refreshed daily, snapshot 2026-10-20.