#833 · Primary category： AI Agents & Automation
AgentCPM-GUI
AgentCPM-GUI: An on-device GUI agent for operating Android apps, enhancing reasoning ability with reinforcement fine-tuning for efficient task execution.
Project last updated：01/11/26
GitHub Stars
1.4K
Forks
134
Contributors
6
License
Apache-2.0
Why we included this project
Most GUI agents are built and tested against clean English web pages, which makes phone-level automation inside real apps a different problem. AgentCPM-GUI is an 8B vision-language model that takes an Android screenshot and returns the next action as compact JSON with a bounding coordinate and a short reasoning note. Its distinguishing strength is Chinese-app coverage: it was fine-tuned on popular titles such as Amap, Dianping, bilibili, and Xiaohongshu, and reinforcement fine-tuning lets it reason before acting on complex tasks. The repo includes the model weights, an evaluation harness, and working inference through Hugging Face transformers or a vLLM server, so you can point it at your own screenshots without much setup. If you are working on on-device UI automation, app testing, or accessibility tooling and your targets are not all English, this is a concrete starting point worth trying.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
openclaw
Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞
hermes-agent
The agent that grows with you
n8n
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.
deepseek-harness
DeepSeek Harness: Everything is a Plugin.
AutoGPT
AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.