PolyWorkBench (Hermes Agent): leaderboard
PolyWorkBench (Beijing Jiaotong University and Tencent, 2026) runs LLM agents through 67 multilingual long-horizon workplace workflows in commerce, knowledge work, legal analysis, localization and manufacturing, scoring the produced artefacts with weighted deterministic rubric checks. Every model runs inside the Hermes Agent harness. Mean Grade in percent from the best of up to three runs, higher is better.
Metric: Mean Grade (%; weighted deterministic rubric checks of the output artefacts, mean over 67 multilingual long-horizon workplace tasks in five domains; best of up to three runs, Pass@1; the Hermes Agent harness, 1,800 s per task). Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 83.7 | #49 |
| 2 | GLM-5.2 | 82.3 | #138 |
| 3 | Claude Opus 4.8 | 80.5 | #56 |
| 4 | Claude Opus 4.7 | 80 | #79 |
| 5 | GLM-5.1 | 79 | #179 |
| 6 | DeepSeek V4 Flash | 75.8 | #258 |
| 7 | Qwen 3.6 27B | 74.2 | #341 |
| 8 | Qwen 3.6 35B A3B | 72.7 | #462 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=polyworkbench-hermes-agent · How It Works · Data refreshed daily, snapshot 2026-11-13.