#6565th of 70 in Agent frameworks and SDKs
mobileagent
GUI agent models and multi-agent code for operating phones, PCs and browsers
- Stars
- 9.3k
- License
- MIT
- Last commit
- Jul 2026
- Language
- Python
8,274 on 1 Apr 20269,321 on 21 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Research repository from Tongyi Lab (Alibaba) collecting the Mobile-Agent series, from v1 to v3.5, plus related work such as PC-Agent, UI-S1 and GUI-Critic-R1. It centers on the GUI-Owl vision-language models (2B to 235B, Instruct and Thinking, built on Qwen3-VL) and multi-agent frameworks that plan, reflect and keep memory while driving mobile, desktop and web interfaces. The README mostly links out to per-project subfolders, papers and hosted demos.
Who it is for: Researchers and engineers building GUI automation agents for phones and desktops
Strengths
- Open GUI-Owl weights in several sizes (2B/4B/8B/32B/235B) on Hugging Face and ModelScope
- Covers mobile, desktop and browser targets, with evaluation code for OSWorld and AndroidWorld
- Each generation (v1 to v3.5, UI-S1, PC-Agent) has its own paper and code folder
- MIT license; hosted demos and a Bailian API exist for trying it without local setup
Weaknesses
- Top-level README has no install or quick-start steps; setup lives in subfolders
- Hosted demos and API depend on Alibaba Cloud services, with docs largely in Chinese
- Larger models (32B, 235B) need substantial GPU memory; requirements are not stated
- No tagged releases; the repo is a collection of separate research projects
What it needs
- Needs Android device or emulator (mobile use), Alibaba Cloud Bailian API (optional hosted inference)
- Models: GUI-Owl-1.5 (2B/4B/8B/32B/235B), GUI-Owl-7B, GUI-Owl-32B, UI-S1-7B
Also in Agent frameworks and SDKs
See all 70| Rank | Project | Score |
|---|---|---|
| 1 | ppt-masterAgent skill that turns documents into natively editable PowerPoint decks | 94 out of 100 |
| 2 | langchainPython framework for building LLM agents and apps from interoperable components | 89 out of 100 |
| 3 | browser-usePython library that lets an LLM agent control a web browser | 89 out of 100 |
| #4 | openai-agents-pythonPython SDK for multi-agent workflows with handoffs, guardrails and tracing | 89 out of 100 |
| #5 | mastraTypeScript framework for building AI agents and workflows | 89 out of 100 |