PolyWorkBench (OpenClaw): leaderboard

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 OpenClaw ReAct-style scaffold with shell, file and Python tools, 1,800 s per task). Source: arxiv.org. Saturation forecast: Around December 2026. 8 models tracked.

Top models

#ModelScore
1GLM-5.285.3
2GPT-5.579.4
3GLM-5.178.3
4Qwen 3.6 27B78.2
5Claude Opus 4.877.8
6Claude Opus 4.770.9
7DeepSeek V4 Flash70.8
8Qwen 3.6 35B A3B67.3

Interactive version: theaggregate.ai/benchmark?slug=polyworkbench-openclaw · How It Works · Data refreshed daily, snapshot 2026-09-29.