CloudOpsBench v1.0: leaderboard
CloudOpsBench v 1.0:45 seeded AWS incident tasks,4 trials each,40 minutes. Native state-based resolution rate, passes/(passes+failures+timeouts), evaluation errors excluded. Model,effort andharness kept distinct.
Metric: Resolution rate (%). Source: cloudopsbench.ai. Saturation forecast: Around January 2028. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 (Max) | 60.6 | #10 (Claude Fable 5.1) |
| 2 | GPT-6 Astra (Max) | 58.9 | #6 (GPT-6 Astra) |
| 3 | Gemini 4 Argon (High) | 57.5 | #9 (Gemini 4 Argon) |
| 4 | Grok 4.7 (High) | 48.3 | #26 (Grok 4.7) |
| 5 | Claude Opus 5.5 (High) | 47.8 | #3 (Claude Opus 5.5) |
| 6 | Claude Fable 5.1 (High) | 43.8 | #10 (Claude Fable 5.1) |
| 7 | Gemini 3.8 Flash (High) | 42.8 | #30 (Gemini 3.8 Flash) |
| 8 | GPT-6 Astra (High) | 42.8 | #6 (GPT-6 Astra) |
| 9 | Grok 4.7 (Medium) | 40 | #26 (Grok 4.7) |
| 10 | Grok 4.6 (Medium) | 36.7 | #29 (Grok 4.6) |
| 11 | Grok 4.6 (High) | 36.1 | #29 (Grok 4.6) |
| 12 | Qwen 3.8 Max (High) | 33.7 | #34 (Qwen 3.8 Max) |
| 13 | DeepSeek V4.1 Flash (High) | 21.8 | #37 (DeepSeek V4.1 Flash) |
| 14 | Kimi K3.1 Pro (High) | 19.1 | #17 (Kimi K3.1 Pro) |
No result here: #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=cloudopsbench-v1-0 · How It Works · Data refreshed daily, snapshot 2026-11-05.