CutVerse: A Compositional GUI Agents Benchmark for Media Post-Production Editing

2026-05-19

AI Summary

This work addresses the lack of systematic evaluation for GUI-based intelligent agents in professional creative workflows—particularly in complex, long-horizon, multimodal interaction tasks such as media post-production. To bridge this gap, the authors introduce the first benchmark specifically designed for media editing, encompassing 186 real-world tasks across seven professional software applications. They also propose a lightweight parser that converts screen recordings and low-level interaction logs into structured action trajectories, enabling composable and extensible evaluation of long-horizon multimodal tasks. Experimental results reveal that current agents achieve only a 36.0% task success rate on this benchmark, highlighting significant deficiencies in their long-horizon reliability and domain-specific planning capabilities.

Abstract

While GUI agents have made significant progress in web navigation and basic operating system tasks, their capabilities in professional creative workflows remain largely underexplored. To bridge this gap, we introduce Cutverse, a benchmark designed to systematically evaluate autonomous GUI agents in realistic media post-production environments. We curate expert demonstrations across 7 professional applications (e.g., Premiere Pro, Photoshop), covering 186 complex, long-horizon tasks grounded in authentic editing workflows, involving dense multimodal interfaces and tightly coupled interaction sequences. To support scalable evaluation, we develop a lightweight parser that transforms raw screen recordings and low-level interaction logs into structured, compositional GUI action trajectories with precise grounding. Extensive evaluations reveal that existing agents achieve only 36.0\% task success on realistic media editing tasks, underscoring the challenges posed by complex, long-horizon media post-production workflows in our benchmark.While current models demonstrate promising spatial grounding, multimodal alignment, and coordinated action execution, they remain limited in long-horizon reliability and domain-specific planning.

Problem

• GUI agents
• media post-production
• benchmark
• long-horizon tasks
• creative workflows

Innovation

• GUI agents
• media post-production
• compositional benchmark
• multimodal interaction
• long-horizon tasks