AutomationBench: leaderboard
AutomationBench API-mode strict task-completion pass rate on the source’s private workflow task set. This board preserves every single-model and effort configuration; default-fallback and named mixed-model pipelines have separate protocol boards.
Metric: Strict Task Pass Rate (%). Source: zapier.com. Saturation forecast: Around April 2028. 109 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Argon (High) | 51.29 | #8 (Gemini 4 Argon) |
| 2 | Gemini 4 Argon (Medium) | 50.08 | #8 (Gemini 4 Argon) |
| 3 | GPT-6 Astra (Max) | 41.4 | #6 (GPT-6 Astra) |
| 4 | GPT-6 Astra (xHigh) | 38.96 | #6 (GPT-6 Astra) |
| 5 | GPT-6 Astra (High) | 37.14 | #6 (GPT-6 Astra) |
| 6 | GPT-6 Astra (Medium) | 34.09 | #6 (GPT-6 Astra) |
| 7 | GPT-6 Sol (xHigh) | 33.2 | #14 (GPT-6 Sol) |
| 8 | GPT-6 Sol (Max) | 32 | #14 (GPT-6 Sol) |
| 9 | GPT-6 Sol (High) | 31.2 | #14 (GPT-6 Sol) |
| 10 | Gemini 3.7 Flash (High) | 30.44 | #23 (Gemini 3.7 Flash) |
| 11 | GPT-6 Astra (Low) | 30.29 | #6 (GPT-6 Astra) |
| 12 | Gemini 3.8 Flash (High) | 29.68 | #21 (Gemini 3.8 Flash) |
| 13 | Gemini 3.8 Flash (Medium) | 29.68 | #21 (Gemini 3.8 Flash) |
| 14 | GPT-5.6 Sol (Max) | 28.77 | #19 (GPT-5.6 Sol) |
| 15 | Claude Opus 5 (Max) | 26.94 | #12 (Claude Opus 5) |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=automationbench · How It Works · Data refreshed daily, snapshot 2026-10-20.