ProSoftArena: Benchmarking Hierarchical Capabilities of Multimodal Agents in Professional Software Environments
This addresses the problem of evaluating AI agents in real-world professional software workflows for researchers and developers, though it is incremental as it extends benchmarking to new domains.
The authors tackled the lack of benchmarks for multimodal agents in professional software environments by introducing ProSoftArena, a benchmark with 436 tasks across 13 applications, and found that even the best agent achieved only a 24.4% success rate on intermediate tasks and failed completely on advanced workflows.
Multimodal agents are making rapid progress on general computer-use tasks, yet existing benchmarks remain largely confined to browsers and basic desktop applications, falling short in professional software workflows that dominate real-world scientific and industrial practice. To close this gap, we introduce ProSoftArena, a benchmark and platform specifically for evaluating multimodal agents in professional software environments. We establish the first capability hierarchy tailored to agent use of professional software and construct a benchmark of 436 realistic work and research tasks spanning 6 disciplines and 13 core professional applications. To ensure reliable and reproducible assessment, we build an executable real-computer environment with an execution-based evaluation framework and uniquely incorporate a human-in-the-loop evaluation paradigm. Extensive experiments show that even the best-performing agent attains only a 24.4\% success rate on L2 tasks and completely fails on L3 multi-software workflow. In-depth analysis further provides valuable insights for addressing current agent limitations and more effective design principles, paving the way to build more capable agents in professional software settings. This project is available at: https://prosoftarena.github.io.