OSUniverse: Benchmark for Multimodal GUI-navigation AI Agents
Mariya Davydova, Daniel Jeffries, Patrick Barker, Arturo Márquez Flores, Sinéad Ryan
TL;DR
OSUniverse addresses the gap in benchmarks for multimodal GUI-navigation by introducing a desktop-focused, extensible benchmark with automated validation. It integrates YAML-defined test cases, modular architecture components (environment, action/observation spaces, runtimes, validators), and a Gemini-based automatic scoring pipeline to handle nondeterminism, enabling scalable progress tracking. The study reports comparative performance across 8 configurations, highlighting that domain-specific GUI-navigation models achieve the strongest, yet still imperfect, results (with total scores often below 50% and stability varying across models). The benchmark’s open-source nature and automated validation framework aim to accelerate development of robust, generalizing GUI-navigation agents and facilitate reproducible progress in real-world desktop tasks.
Abstract
In this paper, we introduce OSUniverse: a benchmark of complex, multimodal desktop-oriented tasks for advanced GUI-navigation AI agents that focuses on ease of use, extensibility, comprehensive coverage of test cases, and automated validation. We divide the tasks in increasing levels of complexity, from basic precision clicking to multistep, multiapplication tests requiring dexterity, precision, and clear thinking from the agent. In version one of the benchmark, presented here, we have calibrated the complexity of the benchmark test cases to ensure that the SOTA (State of the Art) agents (at the time of publication) do not achieve results higher than 50%, while the average white collar worker can perform all these tasks with perfect accuracy. The benchmark can be scored manually, but we also introduce an automated validation mechanism that has an average error rate less than 2%. Therefore, this benchmark presents solid ground for fully automated measuring of progress, capabilities and the effectiveness of GUI-navigation AI agents over the short and medium-term horizon. The source code of the benchmark is available at https://github.com/agentsea/osuniverse.