UI-Mate: Advancing Open-Weight Foundation GUI Agents with In-Context Demonstrations

2026-08-16

AI Summary

This study addresses the scarcity of training data, ambiguous prompting, and unreliable execution in GUI agents by proposing an environment-driven training stack coupled with a contextual demonstration learning mechanism. By integrating a closed-loop data engine, online reinforcement learning, and a unified task verifier, the framework transforms multimodal demonstrations into adaptive subtask flows with real-time replanning capabilities. Accompanied by the newly released OSWorkerBench benchmark, this approach achieves new open-source state-of-the-art performance on evaluations such as OSWorld. Notably, a single demonstration increases the strict success rate of long-horizon tasks from 17.2% to 35.4%, significantly enhancing both the generalization capability and execution reliability of open-weight agents in complex office environments.

Abstract

Foundation GUI agents can automate complex digital tasks, but deployment is hindered by scarce and biased training data, ambiguous prompts, and unreliable execution. Routine workflows rely on user-specific tools and tacit conventions, so unstated instructions can produce arbitrary variations across runs. We present UI-Mate, a foundation GUI agent that integrates an environment-grounded training stack with in-context demonstration learning. UI-Mate makes three contributions: A Scalable Environment-Grounded Training Stack: A closed-loop data engine automates task generation, environment construction, rollout, filtering, capability balancing, SFT, and online RL across massively parallel environments via unified task-verifier bundles. In-Context Demonstration Learning: A mechanism that transforms multimodal demonstrations into flexible subtask-level workflows, follows relevant demonstrated steps, and re-plans from the live interface. OSWorkerBench Benchmark and Insights: A benchmark of 100 long-horizon office tasks across 41 applications that supports instruction-only and demonstration-guided evaluation. Its demonstration resources separate a 33-task self-demo setting, built from successful strong-agent rollouts of the same targets, from a 45-task variant-demo setting, built from human recordings of related but non-identical tasks. Experiments show that UI-Mate-27B sets a new open-weight state of the art on general computer-use benchmarks, scoring 77.0% on OSWorld-Verified and 66.2% on WindowsAgentArena. On OSWorkerBench, it reaches 41.0% strict success and 76.9% progress, outperforming its Qwen3.6-27B base by 17.7 and 24.5 points. On the 33-task self-demo subset, one demonstration raises strict success from 17.2% to 35.4% and progress from 67.9% to 81.1%, substantially improving long-horizon reliability. Project page: https://ui-mate.github.io.

Problem

• GUI Agents
• Training Data Scarcity
• Execution Reliability
• Ambiguous Prompts
• Tacit Conventions

Innovation

• Environment-Grounded Training Stack
• In-Context Demonstration Learning
• Closed-Loop Data Engine
• OSWorkerBench
• GUI Agents