AgentBench

AgentBench is a benchmark that contains a set of deterministic environments used to evaluate the performance of language model agents in various task settings.

Papers using AgentBench (3)

Adapting the Interface, Not the Model: Runtime Harness Adaptation for Deterministic LLM Agents2026AstraFlow: Dataflow-Oriented Reinforcement Learning for Agentic LLMs2026Understanding the Challenges in Iterative Generative Optimization with LLMs2026