OccuBench: leaderboard
Professional-task benchmark using simulated domain tool environments to evaluate LLM agents across occupation-specific workflows.
Metric: Completion rate (self-reported). Source: benchmarklist.com. Status: saturation imminent. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 45.3 |
| 2 | GPT-5.2 | 42.6 |
| 3 | Qwen 3.5 Plus | 41 |
| 4 | Claude Opus 4.6 | 40.1 |
| 5 | MiniMax-M2.7 | 39.4 |
| 6 | GLM-5.1 | 35.7 |
| 7 | Kimi K2.5 | 33.9 |
| 8 | GPT-4o | 28.8 |
| 9 | Qwen 3 Max | 24.8 |
| 10 | Llama 4 Maverick | 22 |
| 11 | Gemma 3 27B | 7 |
| 12 | DeepSeek V3.2 | 1.9 |
| 13 | Gemma 3 12B | 1.5 |
| 14 | Qwen 3.5 4B | 0.2 |
Interactive version: theaggregate.ai/benchmark?slug=occubench · How It Works · Data refreshed daily, snapshot 2026-09-05.