Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Qwen-UI-Agent: Towards Next-Generation Real-World Centric Foundation GUI Agent
| Date | Stars |
|---|---|
| 2026-08-04 | 1890 |
| 2026-08-06 | 1890 |
| 2026-08-13 | 1912 |
| 2026-08-18 | 1917 |
| 2026-08-19 | 1918 |
| 2026-08-20 | 1989 |
| 2026-08-21 | 2111 |
| 2026-08-22 | 2146 |
| 2026-08-23 | 2166 |
| 2026-08-24 | 2198 |
| 2026-08-25 | 2213 |
| 2026-08-26 | 2226 |
| 2026-08-27 | 2237 |
| 2026-08-28 | 2241 |
| 2026-08-29 | 2243 |
| 2026-08-30 | 2246 |
| 2026-08-31 | 2252 |
| 2026-09-01 | 2254 |
| 2026-09-02 | 2257 |
| 2026-09-03 | 2260 |
| 2026-09-04 | 2266 |
| 2026-09-05 | 2268 |
| 2026-09-06 | 2269 |
| 2026-09-07 | 2278 |
| 2026-09-08 | 2283 |
| 2026-09-09 | 2288 |
| 2026-09-10 | 2290 |
| 2026-09-11 | 2294 |
| 2026-09-12 | 2296 |
| 2026-09-13 | 2300 |
| 2026-09-14 | 2304 |
| 2026-09-15 | 2309 |
| 2026-09-16 | 2313 |
| 2026-09-17 | 2315 |
| 2026-09-19 | 2320 |
| 2026-09-20 | 2326 |
| 2026-09-21 | 2328 |
| 2026-09-22 | 2332 |
| 2026-09-23 | 2334 |
| 2026-09-24 | 2355 |
| 2026-09-25 | 2363 |
| 2026-09-26 | 2364 |
| 2026-09-28 | 2366 |
| 2026-09-29 | 2368 |
| 2026-09-30 | 2369 |
| 2026-10-01 | 2370 |
| 2026-10-02 | 2372 |
| 2026-10-03 | 2371 |
| 2026-10-04 | 2373 |
| 2026-10-05 | 2375 |
| 2026-10-06 | 2376 |
| 2026-10-07 | 2380 |
| 2026-10-08 | 2384 |
| 2026-10-09 | 2386 |
| 2026-10-10 | 2388 |
| 2026-10-11 | 2390 |
| 2026-10-12 | 2393 |
| 2026-10-13 | 2395 |
| 2026-10-14 | 2396 |
| 2026-10-15 | 2397 |
| 2026-10-16 | 2398 |
| 2026-10-17 | 2399 |
| 2026-10-18 | 2402 |
| 2026-10-19 | 2406 |
| 2026-10-20 | 2408 |
| 2026-10-21 | 2412 |
| 2026-10-22 | 2414 |
| 2026-10-23 | 2416 |
| 2026-10-24 | 2418 |
| 2026-10-25 | 2419 |
| 2026-10-26 | 2421 |
| 2026-10-27 | 2423 |
| 2026-10-28 | 2424 |
| 2026-10-29 | 2426 |
| 2026-10-30 | 2429 |
| 2026-10-31 | 2428 |
| 2026-11-01 | 2427 |
| 2026-11-04 | 2428 |
Today
+1 stars today
This week
+4 stars this week
This month
+53 stars this month
Momentum
0.0
growth rate 0.17%/day
<p align="center">
<a href="https://tongyi-mai.github.io/Qwen-UI-Agent">
<img src="./Qwen-UI-Agent/assets/img/qwen_mai_title_hd.png" alt="MAI-UI × Qwen-UI-Agent" />
</a>
</p>
<p align="center">
<a href="https://tongyi-mai.github.io/Qwen-UI-Agent/"><img src="https://img.shields.io/badge/🌐%20Website-Qwen--UI--Agent-blue" alt="Qwen-UI-Agent Website" /></a>
<a href="https://tongyi-mai.github.io/Qwen-UI-Agent/Qwen-UI-Agent-Technical-Report.pdf"><img src="https://img.shields.io/badge/📄%20Report-Qwen--UI--Agent-red" alt="Qwen-UI-Agent Technical Report" /></a>
<a href="https://arxiv.org/abs/2607.28227"><img src="https://img.shields.io/badge/📄%20arXiv-Paper-red" alt="arXiv" /></a>
<a href="https://tongyi-mai.github.io/MAI-UI-blog/"><img src="https://img.shields.io/badge/🌐%20Website-MAI%20UI-blue" alt="Website" /></a>
<a href="https://tongyi-mai.github.io/MobileWorld/"><img src="https://img.shields.io/badge/🌐%20Website-MobileWorld-blue" alt="Website" /></a>
</p>
<p align="center">
<img src="./Qwen-UI-Agent/assets/img/maiui_to_qwenuiagent_evolution_v5.png" alt="Evolution from MAI-UI 1.0 to Qwen-UI-Agent" width="850">
</p>
We present **Qwen-UI-Agent**, a real-world-centric foundation GUI agent that unifies **mobile**, **computer**, **browser**, and **DeepSearch** scenarios in a single model. Our core contributions include:
- 📱 **Real-device training & evaluation**: We build a live mobile environment of **100+** physical smartphones covering **150+** apps for task construction, trajectory collection, training, and evaluation, together with **MobileWorld-Real**, a self-built real-device benchmark (**400+** tasks across **100+** apps) that closes the sim-to-real gap.
- ⌨️ **Hybrid GUI + CLI action space with batched actions**: Beyond GUI operations, the agent directly executes Bash commands and emits multiple actions in a single decision. In computer-use tasks, CLI commands and GUI clicks emerge as the two dominant action types, and ~**40%** of action outputs are batched.
- 📈 **Scalable long-horizon online RL**: We run online reinforcement learning over trajectories exceeding **100** steps, with ~**10,000** parallel environments rolling out concurrently to accelerate rollout generation.
- 🔄 **AutoResearch-style data flywheel**: Agents construct tasks, environments, and verifiers, diagnose failures, and plan subsequent iterations, significantly reducing human effort in capability iteration.
- 🔔 **Proactive service & cross-platform workflows**: A harness layer enables the agent to proactively initiate tasks from real-world signals (e.g., a flight-cancellation notification), present decision-ready plans for user confirmation, and execute stateful workflows across mobile and computer.
- 🏆 **State-of-the-art performance**: **82.1%** on MobileWorld, **92.2%** on MobileWorld-Real, **97.5%** on AndroidDaily, **79.5%** on OSWorld-Verified, **40.0%** (partial score) on OSWorld-v2, **73.6%** on WebArena, and **81.5%** on ScreenSpot-Pro.
<p align="center">
<img src="./Qwen-UI-Agent/assets/img/performance_8panel.png" alt="Qwen-UI-Agent performance across eight benchmarks" width="1000">
<br>
<em>Qwen-UI-Agent performance across eight GUI benchmarks</em>
</p>
## Projects
- [**Qwen-UI-Agent**](./Qwen-UI-Agent/) — continuation work of MAI-UI. Visit the [project website](https://tongyi-mai.github.io/Qwen-UI-Agent).
- [**MAI-UI 1.0**](./MAI-UI/) — original MAI-UI repository content. Visit the [project website](https://tongyi-mai.github.io/MAI-UI-blog/).
## 📰 News
* **[2026-07-30]** 🚀 **Introducing Qwen-UI-Agent**: Our follow-up work, [**Qwen-UI-Agent**](https://tongyi-mai.github.io/Qwen-UI-Agent), extends MAI-UI into a real-world centric foundation GUI agent spanning **mobile, computer-use, web, and DeepSearch** environments. It unifies GUI operations with CLI execution in a single action space, emits batched actions per model turn, and is trained with online RL on trajectories exceeding **100 turns** across Excerpt of 13,762 characters
Read on GitHub
28
7
5
5
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:d196f07849a637be, topic:computer-use, topic:gui-agent, desc:gui agent