iOSWorld: A Realistic iOS Environment for Benchmarking Phone Agents with Personalized User Identity and Memory
Lawrence Keunho Jang · Mareks Woodside · Geronimo Carom · Andrew Keunwoo Jang · Jing Yu Koh · Ruslan Salakhutdinov
We build a realistic iOS environment with 26 apps and a rich user identity to benchmark phone agents across 120+ tasks requiring cross-app memory and preference discovery, comparing vision-only, XML-augmented, and leading CUA frameworks.
摘要
A useful phone agent needs to be personally intelligent. It should reason over a user’s identity, history, and preferences as they exist on the device, not just follow isolated instructions in an impersonal sandbox. Existing mobile agent benchmarks lack this kind of personalization. We introduce iOSWorld, the first interactive native iOS simulator benchmark built around a persistent user identity spanning 26 newly built iOS apps. These apps form a controlled ecosystem for one synthetic user, with connected transactions, messages, travel records, social relationships, and financial activity. iOSWorld includes 133 tasks across three increasingly difficult categories. Single-app tasks (27) test one app, multi-app tasks (60) span 2 to 8 apps, and memory and personalization tasks (46) require agents to infer patterns from personal data. We evaluate frontier and open-source computer-use models in both vision-only and privileged vision+XML settings. The best configuration reaches 52% overall but only 37% on multi-app tasks. Privileged vision+XML access improves frontier models by up to 26 percentage points, while smaller models do not benefit from added accessibility-tree input. We release iOSWorld as an open-source benchmark with all apps, seeded data, tasks, rubrics, and evaluation code at https://iosworld.io.