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.
EvoCUA: Evolving Computer Use Agent
| Date | Stars |
|---|---|
| 2026-07-31 | 335 |
| 2026-08-08 | 336 |
| 2026-08-16 | 342 |
| 2026-08-21 | 343 |
| 2026-09-07 | 344 |
| 2026-09-08 | 345 |
| 2026-09-11 | 346 |
| 2026-09-17 | 347 |
| 2026-09-25 | 348 |
| 2026-09-28 | 350 |
| 2026-10-05 | 351 |
| 2026-10-08 | 352 |
| 2026-10-15 | 353 |
| 2026-10-18 | 354 |
| 2026-10-23 | 355 |
| 2026-11-03 | 355 |
| 2026-11-04 | 355 |
Today
— stars today
This week
— stars this week
This month
+4 stars this month
Momentum
0.0
growth rate 0.00%/day
<div align="center"> # EvoCUA: Evolving Computer Use Agent **🥇 #1 Open-Source Model on OSWorld | A General-Purpose Multimodal Model Excelling at Computer Use** [](https://huggingface.co/meituan/EvoCUA-32B-20260105) [](https://huggingface.co/meituan/EvoCUA-8B-20260105) [](https://os-world.github.io/) [](./LICENSE) [English](./README.md) | [中文](./README_CN.md) <img src="assets/images/osworld_leaderboard.png" width="700" alt="OSWorld Leaderboard"> **🥇 #1 Open-Source Model on OSWorld Leaderboard (Jan 2026)** </div> --- ## 📢 Updates - **2026.03.31**: EvoCUA-32B achieves **56.48%** on WindowsAgentArena (WAA), surpassing the base model Qwen3-VL-32B-Thinking (42.9%) by ~13.6 points and UI-TARS-2 (50.6%) by ~6 points — demonstrating strong zero-shot cross-OS generalization 🆕 - **2026.03.31**: Independent safety study by Yoshua Bengio & Dawn Song's teams ([arXiv:2602.08235](https://arxiv.org/abs/2602.08235)) shows **EvoCUA-32B has the lowest unintended-behavior rate (35.0%)** among all tested CUAs — the safest agent! 🆕 - **2026.01.23**: EvoCUA ranked **#1** on [Hugging Face Daily Papers](https://huggingface.co/papers/2601.15876) 🏆 - **2026.01.22**: Released [EvoCUA Technical Report](./tech_report.pdf) 📄 - **2026.01.13**: Released [EvoCUA-8B-20260105](https://huggingface.co/meituan/EvoCUA-8B-20260105) — achieves **46.1%** on OSWorld, **competitive with 72B-level models using fewer parameters!** - **2026.01.05**: Released [EvoCUA-32B-20260105](https://huggingface.co/meituan/EvoCUA-32B-20260105) with **56.7%** on OSWorld, achieving **#1** among open-source models 🥇 --- ## 🌟 Highlights - 🥇 **#1 Open-Source Model on OSWorld**: Achieves **56.7%** task completion rate, **#1 among all open-source models** - 📈 **Significant Improvements**: +11.7% over OpenCUA-72B (45.0%→56.7%), +15.1% over Qwen3-VL thinking (41.6%→56.7%), with fewer parameters and half the steps - 🖥️ **End-to-End Multi-Turn Automation**: Operates Chrome, Excel, PowerPoint, VSCode and more through screenshots and natural language instructions - 🧠 **Novel Training Method**: Our data synthesis and training approach consistently improves Computer Use capability across multiple open-source VLMs without degrading general performance --- ## 📊 Performance Comparison | Rank | Model | Open/Closed | Type | Max Steps | Score | |------|-------|-------------|------|-----------|-------| | 1 | Claude-sonnet-4-5 | 🔒 Closed | General | 100 | 62.9% | | 2 | Seed-1.8 | 🔒 Closed | General | 100 | 61.9% | | 3 | Claude-sonnet-4-5 | 🔒 Closed | General | 50 | 58.1% | | **4** | **EvoCUA-20260105 (Ours)** | **🟢 Open** | **General** | **50** | **56.7% 🥇** | | 5 | DeepMiner-Mano-72B | 🔒 Closed | Specialized | 100 | 53.9% | | 6 | UI-TARS-2-2509 | 🔒 Closed | General | 100 | 53.1% | | 7 | EvoCUA (Previous Version) | 🔒 Closed | General | 50 | 50.3% | | **8** | **EvoCUA-8B-20260105 (Ours)** | **🟢 Open** | **General** | **50** | **46.1%** | | 9 | OpenCUA-72B | 🟢 Open | Specialized | 100 | 45.0% | | ... | ... | ... | ... | ... | ... | | 13 | Qwen3-VL-Flash | 🔒 Closed | General | 100 | 41.6% | > EvoCUA is **#1 among all open-source models**, achieving competitive results with only **50 steps**. Human-level performance remains significantly higher, indicating substantial room for improvement. ### Zero-shot Cross-OS Control (WindowsAgentArena) We evaluated EvoCUA on [WindowsAgentArena (WAA)](https://microsoft.github.io/WindowsAgentArena/) to test generalization from the Linux-based training environment to a wholly different OS platform. As shown below, EvoCUA-32B reaches **56.48%**, surpassing the base model Qwen3-VL-32B-Thinking (42.9%) by ~13.6 points and the leading frontier GUI agent UI-TARS-
Excerpt of 10,292 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:744949c09c3227a4, topic:foundation-models, readme:technical report
matched fp:744949c09c3227a4, topic:vision-language-model, readme:multimodal