CUADebug: Diagnosing and Repairing Computer-Use Agent Failures Paper • 2608.02643 • Published Jul 31 • 3
Business Arena: Benchmarking LLM Agents in a Realistic Marketplace Paper • 2608.08621 • Published Aug 9 • 22
OSReward: Instituting Standardized Evaluation for Cross-Platform Computer-Use Reward Models Paper • 2607.28609 • Published Jul 30 • 73