AutomationBench (Overall)
Agents · 2026-04-20
Overall pass rate on Zapier's end-to-end business workflow automation benchmark.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Opus 5 | 25.1 |
| Gemini 3.6 Flash | 19.8 |
| GPT-5.6 Sol | 18.1 |
| Claude Fable 5 | 17.4 |
| Opus 4.8 | 17.2 |
| Opus 4.7 | 16.3 |
| GPT-5.6 Terra | 15.2 |
| GPT-5.6 Luna | 14.9 |
| Gemini 3.5 Flash | 14.5 |
| Sonnet 5 | 14.2 |
| GLM-5.2 | 14.0 |
| GPT-5.5 | 12.9 |
| Gemini 3.5 Flash-Lite | 10.8 |
| Gemini 3.1 Pro Preview | 9.6 |
| Sonnet 4.6 | 8.7 |
| GPT-5.4 | 7.6 |
| MiniMax M3 | 6.8 |
| GLM-5.1 | 6.2 |
| Kimi K2.7 Code | 4.9 |
| Kimi K2.6 | 4.9 |
| Qwen3.6 Plus (2026-04-02) | 4.3 |
| Qwen3.7-Plus | 3.7 |
| Gemma 4 31B IT | 1.7 |
| MiniMax M2.7 | 1.7 |
| Haiku 4.5 | 1.5 |
| gpt-oss-120b | 0.6 |