Alibaba's Qwen-UI-Agent Surpasses GPT-4o in Multi-Platform GUI Benchmarks
千问大模型 · wechat · 2026-08-20
Alibaba's Qwen team released Qwen-UI-Agent, a real-world-centric GUI agent foundation model covering mobile, desktop, and web environments. It achieves SOTA performance on benchmarks like MobileWorld, OSWorld, and WebArena, outperforming GPT-4o and Claude Opus in several metrics. Key features include training/eval on 100+ real devices, integrated safety boundaries, hybrid GUI/CLI action spaces with batching, online RL for long trajectories, and proactive 'Harness' capabilities. The technical report demonstrates its prowess in cross-platform workflows like cross-device file management and long-horizon research report generation.
Related event: Alibaba Releases Qwen-UI-Agent, Topping 5 of 6 GUI Benchmarks(2 posts)→
More from Embodied
- Professor recruiting one PhD student for Fall 2027 on agentic wearables — jasonwuishere · 2026-10-17
- Sandbar's Stream Ring nails whisper-quiet dictation across iOS and Mac in demo — thesephist · 2026-10-17
- RAX open-sourced: frozen coding agents learn robot tasks via online exploration, zero demos — du_yilun · 2026-10-17
- U. Tokyo's TwinX robot: one wire links two wheeled-legged bots for morphing and teamwork — LexiLove · 2026-10-17
- Snapchat Specs pop-up in LA: tech is cool but the $2k glasses 'look ridiculous' — SimonBalmain · 2026-10-17
- IEEE profiles roboticist Peter Stone: robot soccer, service robots and being a generalist — PeterStone_TX · 2026-10-17