{"slug": "qwen-ui-agent", "title": "Qwen UI Agent", "summary": "Alibaba's MAI-UI Team released Qwen-UI-Agent, a foundation GUI agent that navigates mobile, desktop, and web interfaces to complete real-world tasks, achieving 89.9 on Tau2-Bench and 64.1 on BrowseComp, outperforming Qwen3.5-27B and other models. The model preserves general reasoning and instruction-following capabilities while adding GUI-specific skills, as detailed in the Qwen-UI-Agent Technical Report.", "body_md": "### Mobile GUI Use\n\nOptimized for everyday tasks in real-device mobile GUI use, the model navigates changing Android apps, accounts, content, and network conditions to search, compare, schedule, shop, and coordinate reliably.\n\nTowards Next-Generation Real-World Centric Foundation GUI Agent\n\nOne agent that thinks, searches, and acts across mobile, desktop, and the web to complete real-world, long-horizon tasks.\n\nBuilt to complete real work across GUI interfaces.\n\nOptimized for everyday tasks in real-device mobile GUI use, the model navigates changing Android apps, accounts, content, and network conditions to search, compare, schedule, shop, and coordinate reliably.\n\n**†** Author-reproduced result: the baseline was independently evaluated in the authors’ environment rather than copied from the model provider’s report.\n\nReal-world tasks demand more than interface interaction—they also require knowledge, multimodal reasoning, instruction following, and tool use. Qwen-UI-Agent gains strong GUI capabilities without becoming a narrow GUI-only model, preserving the base model’s general reasoning and agentic strengths for broader tasks.\n\nMultimodal understanding, knowledge, mathematics, and instruction following.\n\nSwipe horizontally to compare all models →\n\n| Benchmark | Qwen-UI-Agent | Qwen3.5-27B | UI-Venus 30B-A3B | GUI-Owl 32B | OpenCUA-72B |\n|---|---|---|---|---|---|\n| MMMU-Pro | 72.4 | 73.5 | 32.4 | 39.5 | 31.0 |\n| RealWorldQA | 83.1 | 83.1 | 75.3 | 76.7 | 66.4 |\n| CharXiv-RQ | 77.7 | 76.8 | 44.7 | 50.9 | 39.6 |\n| MathVision | 82.8 | 82.0 | 36.8 | 50.6 | 26.6 |\n| AI2D_TEST | 91.1 | 91.9 | 84.3 | 84.8 | 78.9 |\n| MMLU-Pro | 86.5 | 86.0 | 65.6 | 73.9 | 58.8 |\n| IFEval (prompt-level strict) | 90.2 | 90.4 | 81.3 | 84.5 | 70.6 |\n\nTool use, terminal tasks, multi-turn service work, coding workflows, and deep-research retrieval on BrowseComp (BC) and BrowseComp-ZH (BC-ZH).\n\nSwipe horizontally to compare all models →\n\n| Benchmark | Qwen-UI-Agent | Qwen3.5-27B | UI-Venus 30B-A3B | GUI-Owl 32B | OpenCUA-72B |\n|---|---|---|---|---|---|\n| Tau2-Bench | 89.9 | 89.2 | 22.7 | 6.1 | 14.4 |\n| Terminal-Bench 2.0 · Avg 5 | 50.1 | 41.1 | 3.2 | 0.0 | 9.0 |\n| Claw-Eval · Avg 3 | 73.5 | 66.9 | 30.6 | 29.6 | 26.4 |\n| Claw-Eval · Pass@3 | 51.8 | 41.2 | 5.5 | 5.5 | 0.5 |\n| BFCL-v4 | 74.2 | 71.3 | 19.8 | 32.7 | 28.3 |\n| SkillsBench · Avg 5 | 28.0 | 24.9 | 0.5 | 0.3 | 0.0 |\n| QwenClawBench · Avg 3 | 44.2 | 48.5 | 6.4 | 5.1 | 11.4 |\n| BrowseComp (BC) | 64.1 | 61.0 | — | — | — |\n| BrowseComp-ZH (BC-ZH) | 75.0 | 62.1 | — | — | — |\n\nAll scores were independently reproduced in the authors' evaluation environment. Some harness, judge, simulator, runtime, or task-subset settings differ from official evaluations and are documented in the technical report.\n\nRecipe research + e-shopping\n\nI’m planning to make “passion-fruit sour-soup beef” tonight. Search Douyin for the most-saved photo-and-text post, save it, and remember the ingredients I need to prepare. Then, in the Hema app, purchase all the ingredients mentioned in the post—excluding seasonings—select delivery for 18:45 today, and place the order.\n\nComplete everyday workflows across changing apps on physical mobile devices.\n\nThe agent first extracts a recipe and its ingredients from Douyin, then carries that information into Hema to complete a time-constrained grocery order.\n\n```\n@misc{qwenuiagent2026,\n  title  = {Qwen-UI-Agent Technical Report:\n            Toward Next-Generation Real-World-Centric\n            Foundation GUI Agents},\n  author = {MAI-UI Team},\n  year   = {2026},\n  note   = {Alibaba Token Hub}\n}\n```\n\n", "url": "https://wpnews.pro/news/qwen-ui-agent", "canonical_source": "https://tongyi-mai.github.io/Qwen-UI-Agent/", "published_at": "2026-08-03 01:44:11+00:00", "updated_at": "2026-08-03 01:52:30.842230+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-agents", "ai-research", "ai-products"], "entities": ["Alibaba", "MAI-UI Team", "Qwen-UI-Agent", "Qwen3.5-27B", "UI-Venus 30B-A3B", "GUI-Owl 32B", "OpenCUA-72B", "Tau2-Bench"], "alternates": {"html": "https://wpnews.pro/news/qwen-ui-agent", "markdown": "https://wpnews.pro/news/qwen-ui-agent.md", "text": "https://wpnews.pro/news/qwen-ui-agent.txt", "jsonld": "https://wpnews.pro/news/qwen-ui-agent.jsonld"}}