让大模型在预算内智能调用工具，避免超支。
Budget-Constrained Agentic Large Language Models: Intention-Based Planning for Costly Tool Use
- 用分层世界模型预判工具使用和风险成本
- 在限定预算下任务成功率显著高于基线
- 适应价格波动和预算变化，适合实用部署
我们研究了预算约束下的工具增强型智能体，即大语言模型需在严格金钱预算下通过调用外部工具完成多步任务。该场景被形式化为上下文空间中的序贯决策问题，面临工具执行有价且随机、状态动作空间巨大、结果方差高、探索成本过高等挑战。为应对这些问题，我们提出 INTENT 框架，在推理时利用意图感知的分层世界模型，预估未来工具使用、风险校准的成本，并在线指导决策。在 cost-augmented StableToolBench 上，INTENT 严格保证硬性预算可行性，同时显著提升任务成功率，并在工具价格变动和预算调整等动态环境下保持鲁棒性。
原文摘要 · Abstract (English)
We study budget-constrained tool-augmented agents, where a large language model must solve multi-step tasks by invoking external tools under a strict monetary budget. We formalize this setting as sequential decision making in context space with priced and stochastic tool executions, making direct planning intractable due to massive state-action spaces, high variance of outcomes and prohibitive exploration cost. To address these challenges, we propose INTENT, an inference-time planning framework that leverages an intention-aware hierarchical world model to anticipate future tool usage, risk-calibrated cost, and guide decisions online. Across cost-augmented StableToolBench, INTENT strictly enforces hard budget feasibility while substantially improving task success over baselines, and remains robust under dynamic market shifts such as tool price changes and varying budgets.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。