AgentBench

AgentBench is a benchmark designed to evaluate the performance and failure modes of agentic AI systems operating continuously in production environments.

Papers using AgentBench (11)

Agentbench: Evaluating Llms As Agents2023 · 53 citesLearning to Control LLM Agent Harnesses with Offline Reinforcement Learning2026ORAgentBench: Can LLM Agents Solve Challenging Operations Research Tasks End to End?2026Adapting the Interface, Not the Model: Runtime Harness Adaptation for Deterministic LLM Agents2026What Do Agents Learn From Trajectory-sft: Semantics Or Interfaces?2026SAGE-32B: Agentic Reasoning Via Iterative Distillation2026Evaluating Agentic AI In The Wild: Failure Modes, Drift Patterns, And A Production Evaluation Framework2026AstraFlow: Dataflow-Oriented Reinforcement Learning for Agentic LLMs2026Understanding the Challenges in Iterative Generative Optimization with LLMs20266GAgentGym: Tool Use, Data Synthesis, and Agentic Learning for Network Management2026Enhancing the General Agent Capabilities of Low-Parameter LLMs through Tuning and Multi-Branch Reasoning2024 · 1 cites