OpenGVLab/ScaleCUA-7B
ScaleCUA-7B by OpenGVLab is a 7 billion parameter vision-language model designed for automating interactions with graphical user interfaces across multiple operating systems. It is trained on a large-scale, diverse corpus spanning 6 operating systems and 3 task domains, enabling seamless cross-platform operation. The model excels at GUI understanding, grounding, and task completion, achieving state-of-the-art performance on benchmarks like MMBench-GUI L1-Hard (94.4%), OSWorld-G (60.6%), and WebArena-Lite-v2 (47.4%). Its primary use case is developing general-purpose computer-use agents capable of interacting with various GUIs.
Loading preview...
ScaleCUA-7B: Cross-Platform Computer Use Agent
ScaleCUA-7B is a 7 billion parameter vision-language model developed by OpenGVLab, specifically engineered to automate interactions with graphical user interfaces (GUIs) across diverse operating systems. Unlike many existing computer use agents built on proprietary models or datasets, ScaleCUA-7B is trained on a novel, large-scale open-source corpus collected via a dual-loop interactive pipeline involving automated agents and human experts. This dataset covers 6 operating systems and 3 task domains, providing a robust foundation for cross-platform functionality.
Key Capabilities
- Cross-Platform GUI Automation: Operates seamlessly across heterogeneous platforms, including desktops, mobile devices, and web browsers.
- Enhanced GUI Understanding & Grounding: Demonstrates strong performance in interpreting visual interfaces and grounding actions.
- State-of-the-Art Performance: Achieves high success rates on challenging benchmarks, including 94.4% on MMBench-GUI L1-Hard, 60.6% on OSWorld-G, and 47.4% on WebArena-Lite-v2.
- Dual Action Modes: Supports a "Direct Action Mode" for immediate, executable GUI grounding and a "Reasoned Action Mode" for complex, multi-step tasks requiring planning and sequential actions.
Good For
- Developing autonomous agents for automating tasks across various operating systems.
- Research and development in computer vision, natural language processing, and human-computer interaction.
- Applications requiring precise GUI interaction and task completion in diverse digital environments.