OccuBench (Clean Environment): leaderboard

OccuBench (2026) evaluates agents on 382 professional tool-use tasks from 100 occupational scenarios in 10 industries, against environments simulated by Gemini-3-Flash-Preview, each trajectory passed or failed by a rubric verifier; thinking mode on, high effort where configurable. This board is the completion rate over all tasks in the clean environment, in percent; higher is better.

Metric: Completion rate (%) over all 382 tasks in the clean environment (E0); 382 tool-use task instances from 100 professional scenarios, the environment simulated by Gemini-3-Flash-Preview and each trajectory passed or failed by a rubric verifier; thinking mode on, high reasoning effort where configurable; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 15 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-5.2High79.6#185
2Gemini 3.1 Pro (Preview)72.3#99
3Claude Opus 4.6High71.5#113
4Qwen 3.5 PlusThinking69.9#242
5DeepSeek V3.2Thinking69.6#419
6Claude Opus 4.5Thinking65.2#160
7Claude Sonnet 4.5Thinking64.9#261
8Claude Sonnet 4.6Thinking64.4#173
9Kimi K2.5Thinking64.1#266
10GLM-5Thinking62.6#269
11Claude Opus 4Thinking61.3#326
12Gemini 3.1 Flash LiteThinking61.3#371
13Qwen 3.5 FlashThinking59.7#441
14MiniMax-M2.753.9#539
15Claude Sonnet 4Thinking53.4#410

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=occubench-clean-environment · How It Works · Data refreshed daily, snapshot 2026-11-21.