CostBench: Evaluating Multi-Turn Cost-Optimal Planning and Adaptation in Dynamic Environments for LLM Tool-Use Agents

TL;DR

CostBench is a new benchmark for evaluating LLM agents' ability to plan cost-effectively and adapt to dynamic, unpredictable environments, highlighting current limitations in cost-aware reasoning.

Contribution

The paper introduces CostBench, a scalable, cost-centric benchmark with dynamic challenges to assess and improve LLM agents' economic reasoning and replanning skills.

Findings

Agents often fail to find cost-optimal solutions in static settings.

GPT-5 achieves less than 75% accuracy on complex tasks.

Performance drops by around 40% under dynamic conditions.

Abstract

Current evaluations of Large Language Model (LLM) agents primarily emphasize task completion, often overlooking resource efficiency and adaptability. This neglects a crucial capability: agents' ability to devise and adjust cost-optimal plans in response to changing environments. To bridge this gap, we introduce CostBench, a scalable, cost-centric benchmark designed to evaluate agents' economic reasoning and replanning abilities. Situated in the travel-planning domain, CostBench comprises tasks solvable via multiple sequences of atomic and composite tools with diverse, customizable costs. It also supports four types of dynamic blocking events, such as tool failures and cost changes, to simulate real-world unpredictability and necessitate agents to adapt in real time. Evaluating leading open-sourced and proprietary models on CostBench reveals a substantial gap in cost-aware planning: agents frequently fail to identify cost-optimal solutions in static settings, with even GPT-5 achieving less than 75% exact match rate on the hardest tasks, and performance further dropping by around 40% under dynamic conditions. By diagnosing these weaknesses, CostBench lays the groundwork for developing future agents that are both economically rational and robust.