VitaBench: Benchmarking LLM Agents with Versatile Interactive Tasks in Real-world Applications

Wei He, Yueqing Sun, Hongyan Hao +13

As LLM-based agents are increasingly deployed in real-life scenarios, existing benchmarks fail to capture their inherent complexity of handling extensive information, leveraging di…