Agent Planning Benchmark: A Diagnostic Framework for Planning Capabilities in LLM Agents
The Agent Planning Benchmark (APB) is a diagnostic framework designed to evaluate Large Language Model (LLM) agents' planning abilities before they execute tasks. It systematically tests different planning styles across 4,209 cases to pinpoint exactly where an agent fails—whether it's in logic, tool use, or long-term reasoning. This helps researchers understand and fix specific planning weaknesses in AI agents.