# AutomationBench — AI model rankings

Tests whether the AI can run real multi-step business workflows — the kind of end-to-end office processes companies want to automate — from start to finish. Higher is better.

19 tracked models have a published AutomationBench score. Higher is better. Scores come from published lab reports and benchmark sources; recorded sources appear beside the scores.

## Ranking

| Rank | Model | Developer | Score | Source | Released |
| --- | --- | --- | --- | --- | --- |
| 1 | DeepSeek-V4.1-Flash-1017 | DeepSeek | 60.5% | Lab | Oct 17 2026 |
| 2 | Mistral Large 4 | Mistral | 59.9% | Lab | Oct 6 2026 |
| 3 | DeepSeek-V4.1-Flash | DeepSeek | 54.8% | Lab | Sep 10 2026 |
| 4 | Gemini 4 Argon | Google | 51.3% | Lab | Sep 30 2026 |
| 5 | Qwen3.8-Max-0902 | Qwen | 50.8% | Lab | Sep 2 2026 |
| 6 | Muse Spark 1.3 | Meta | 49.4% | Lab | Sep 2 2026 |
| 7 | GLM-5.3-Flash | Z.ai | 48.8% | Lab | Aug 26 2026 |
| 8 | GLM-5.3 | Z.ai | 48.2% | Lab | Aug 14 2026 |
| 9 | Muse Wonder Full | Meta | 45.4% | Lab | Oct 12 2026 |
| 10 | GPT-6 Astra | OpenAI | 41.4% | Lab | Sep 3 2026 |
| 11 | Claude Opus 5.5 | Anthropic | 40% | Lab | Sep 22 2026 |
| 12 | Muse Wonder | Meta | 39.1% | Lab | Oct 12 2026 |
| 13 | GPT-6.1 Sol | OpenAI | 36.2% | Lab | Sep 29 2026 |
| 14 | GPT-6 Sol | OpenAI | 33.2% | Lab | Sep 22 2026 |
| 15 | DeepSeek-V4-Pro-0813 | DeepSeek | 31.8% | Lab | Aug 13 2026 |
| 16 | Claude Fable 5.1 | Anthropic | 31.4% | Lab | Sep 1 2026 |
| 17 | Gemini 3.7 Flash | Google | 30.4% | Lab | Aug 13 2026 |
| 18 | Claude Opus 5 | Anthropic | 26.9% | Lab | Jul 24 2026 |
| 19 | DeepSeek-V4-Flash-0731 | DeepSeek | 25.1% | Lab | Jul 31 2026 |


---

Canonical page: https://aireleasetracker.com/benchmark/automationbench
Site index: https://aireleasetracker.com/llms.txt
Source: AI Release Tracker (https://aireleasetracker.com). Most benchmark scores come from lab launch material; gathered results identify the leaderboard that published them.