AutomationBench-AA - Strict: leaderboard
Artificial Analysis evaluates agents carrying out workplace objectives subject to guardrails. The headline score adjusts objective completion for violations; the strict board requires full task success. Both are percentages from the complete public model roster.
Metric: Strict Task Success (%). Source: artificialanalysis.ai. Saturation forecast: Around June 2027. 243 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Carbon (High) | 55.25 | #1 (Gemini 4 Carbon) |
| 2 | Gemini 4 Argon (High) | 53.12 | #9 (Gemini 4 Argon) |
| 3 | Kimi K3.1 Pro | 45.97 | #17 |
| 4 | Muse Wonder Full (Max) | 45.36 | #12 (Muse Wonder Full) |
| 5 | GPT-6 Astra (Max) | 41.55 | #6 (GPT-6 Astra) |
| 6 | GPT-6 Astra (xHigh) | 39.42 | #6 (GPT-6 Astra) |
| 7 | Muse Wonder (Max) | 39.12 | #18 (Muse Wonder) |
| 8 | MiniMax-M3.1-Flash | 38.05 | #52 |
| 9 | Grok 4.7 (xHigh) | 37.75 | #26 (Grok 4.7) |
| 10 | Grok 4.7 (High) | 37.75 | #26 (Grok 4.7) |
| 11 | GPT-6 Astra (High) | 37.44 | #6 (GPT-6 Astra) |
| 12 | GPT-6.1 Sol (xHigh) | 37.14 | #7 (GPT-6.1 Sol) |
| 13 | GPT-6.1 Sol (Max) | 36.99 | #7 (GPT-6.1 Sol) |
| 14 | DeepSeek V4.1 Flash (1017) (Max) | 36.68 | #27 (DeepSeek V4.1 Flash (1017)) |
| 15 | DeepSeek V4.1 Flash (1017) (High) | 35.77 | #27 (DeepSeek V4.1 Flash (1017)) |
No result here: #3 Claude Opus 5.5, #10 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=automationbench-aa-strict · How It Works · Data refreshed daily, snapshot 2026-11-05.