Claw-Anything - CLI and GUI Tasks: leaderboard

Claw-Anything (2026) benchmarks always-on personal assistant agents in a simulated digital world with months of user activity, interdependent backend services and CLI and GUI devices, all run in the OpenHarness agent scaffold. This board is pass@1 in percent on the tasks that need both the command line and a GUI phone device; higher is better.

Metric: Pass@1 (%) on the 50 tasks that need both the CLI and the GUI device, over the 200 Claw-Anything tasks (simulated months of user activity, interdependent backend services, CLI Linux and GUI Android devices, reactive and proactive tasks), every model in the OpenHarness agent scaffold, three independent runs, rubric checks plus a Claude Sonnet 4.5 judge giving a pass label and a soft score; higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 8 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 3.6 27B18#351
2GPT-5.517.3#54
3GLM-5.116#198
4MiniMax-M2.711.3#539
5Kimi K2.69.3#192
6Qwen 3.5 27B9.3#375
7Claude Opus 4.77.3#85
8Claude Sonnet 4.56#261

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

Interactive version: theaggregate.ai/benchmark?slug=claw-anything-cli-and-gui-tasks · How It Works · Data refreshed daily, snapshot 2026-11-21.