macOSWorld: A Multilingual Interactive Benchmark for GUI Agents

TL;DR

macOSWorld is the first comprehensive multilingual benchmark for GUI agents on macOS, evaluating their performance across diverse tasks, languages, and safety vulnerabilities, revealing significant gaps in current models.

Contribution

Introduces macOSWorld, a novel benchmark with 202 multilingual GUI tasks on macOS, including safety evaluation, to advance GUI agent research in a new OS domain.

Findings

Proprietary agents outperform open-source models with over 30% success.

Open-source models lag below 5% success rate.

Multilingual performance drops notably in Arabic, with 28.8% degradation.

Abstract

Graphical User Interface (GUI) agents show promising capabilities for automating computer-use tasks and facilitating accessibility, but existing interactive benchmarks are mostly English-only, covering web-use or Windows, Linux, and Android environments, but not macOS. macOS is a major OS with distinctive GUI patterns and exclusive applications. To bridge the gaps, we present macOSWorld, the first comprehensive benchmark for evaluating GUI agents on macOS. macOSWorld features 202 multilingual interactive tasks across 30 applications (28 macOS-exclusive), with task instructions and OS interfaces offered in 5 languages (English, Chinese, Arabic, Japanese, and Russian). As GUI agents are shown to be vulnerable to deception attacks, macOSWorld also includes a dedicated safety benchmarking subset. Our evaluation on six GUI agents reveals a dramatic gap: proprietary computer-use agents lead at above 30% success rate, while open-source lightweight research models lag at below 5\%, highlighting the need for macOS domain adaptation. Multilingual benchmarks also expose common weaknesses, especially in Arabic, with a 28.8% average degradation compared to English. Results from safety benchmarking also highlight that deception attacks are more general and demand immediate attention. Project page: https://macos-world.github.io.