September 23, 2026
Qwen gives smartphones end-to-end tasks: Mobile-Use Agent completes 90%
On September 23, Qwen reported that Mobile-Use Agent completes 90% of tasks end to end. Qwen Intelligence added two more AI agents: a task planner and a generator that turns a single phrase into finished content. The generator creates an image in 3 seconds.
Qwen
@alibaba_qwen
Introducing Qwen Intelligence: personal intelligence is now available to everyone. 📱✨ It launches with three best-in-class AI agents: 🥳 - Mobile Planner Agent: plans, breaks down complex tasks and coordinates them. No. 1 in MobilePA-Bench, MobilePA-Bench Business and Memory. - Mobile-Use Agent: completes tasks through APIs first and, when needed, through the interface. MobileWorld 82.1, MobileWorld-Real 92.2, AndroidDaily 97.2, 90% end-to-end task success. - Mobile Creative Agent: turns a single phrase into finished content. An image is generated in 3 seconds, roughly twice as fast as leading alternatives. We are also releasing our benchmark suite: MobilePA-Bench, MobileWorld, MobileWorld-Real and MobileWorld-Safety. They test planning, cross-app operation, work on real devices and safety. 🔗 Learn more about the agents: - Qwen Intelligence official website: https://www.qwenintelligence.com - Mobile Planner Agent: https://github.com/Tongyi-MAI/Qwen-Planner-Agent - Mobile-Use Agent: https://tongyi-mai.github.io/Qwen-UI-Agent/ - Mobile Creative Agent: https://arxiv.org/abs/2608.16887 🔗 Open benchmark suite: - MobilePA-Bench: https://tongyi-mai.github.io/MobilePA-Bench/ - MobileWorld (GitHub): https://github.com/Tongyi-MAI/MobileWorld - Leaderboard: https://tongyi-mai.github.io/MobileWorld/#leaderboard
· 128.8K views
Previously, an agent for smartphones had to act through an app's interface. Mobile-Use Agent calls the API first and switches to the interface when needed.
Testing in practice. Qwen released MobilePA-Bench, MobileWorld, MobileWorld-Real and MobileWorld-Safety. They test planning, cross-app operation, task completion on real devices and safety.
Open benchmarks provide the next comparison point for mobile AI agents.
