Benchmarks/agents/OSWorld-G

OSWorld-G

OSWorld-G (Grounding) evaluates screenshot grounding accuracy for OS automation tasks.

Progress Over Time

Interactive timeline showing model performance evolution on OSWorld-G

State-of-the-art frontier
Open
Proprietary

OSWorld-G Leaderboard

1 models • 0 verified
ContextCostLicense
1
Alibaba Cloud / Qwen Team
Qwen3 VL 235B A22B Thinking
Alibaba Cloud / Qwen Team
0.683236B262K
$0.45
$3.49
Notice missing or incorrect data?Start an Issue discussion→

FAQ

Common questions about OSWorld-G

OSWorld-G (Grounding) evaluates screenshot grounding accuracy for OS automation tasks.
The OSWorld-G leaderboard ranks 1 AI models based on their performance on this benchmark. Currently, Qwen3 VL 235B A22B Thinking by Alibaba Cloud / Qwen Team leads with a score of 0.683. The average score across all models is 0.683.
The highest OSWorld-G score is 0.683, achieved by Qwen3 VL 235B A22B Thinking from Alibaba Cloud / Qwen Team.
1 models have been evaluated on the OSWorld-G benchmark, with 0 verified results and 1 self-reported results.
OSWorld-G is categorized under agents, grounding, multimodal, and vision. The benchmark evaluates image models.