Ares: Adaptive Reasoning Effort Selection for Efficient LLM Agents

2026-03-09

AI Summary

This work addresses the high computational cost of large language model (LLM) agents in multi-step tasks, where sustained high-intensity reasoning leads to excessive token consumption, while static or random selection of reasoning intensity fails to balance efficiency and performance. To this end, we propose Ares, a novel framework that enables step-wise adaptive selection of reasoning intensity for the first time in multi-step agent tasks. Ares employs a lightweight routing module that dynamically predicts the minimal required reasoning intensity at each step based on historical interactions. The approach is plug-and-play, compatible with any LLM supporting multi-level reasoning, and leverages synthetic data generation and fine-tuning for intensity annotation. Evaluated on TAU-Bench, BrowseComp-Plus, and WebArena, Ares reduces reasoning token usage by up to 52.7% with negligible degradation in task success rates.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Modern agents powered by thinking LLMs achieve high accuracy through long chain-of-thought reasoning but incur substantial inference costs. While many LLMs now support configurable reasoning levels (e.g., high/medium/low), static strategies are often ineffective: using low-effort modes at every step leads to significant performance degradation, while random selection fails to preserve accuracy or provide meaningful cost reduction. However, agents should reserve high reasoning effort for difficult steps like navigating complex website structures, while using lower-effort modes for simpler steps like opening a target URL. In this paper, we propose Ares, a framework for per-step dynamic reasoning effort selection tailored for multi-step agent tasks. Ares employs a lightweight router to predict the lowest appropriate reasoning level for each step based on the interaction history. To train this router, we develop a data generation pipeline that identifies the minimum reasoning effort required for successful step completion. We then fine-tune the router to predict these levels, enabling plug-and-play integration for any LLM agents. We evaluate Ares on a diverse set of agent tasks, including TAU-Bench for tool use agents, BrowseComp-Plus for deep-research agents, and WebArena for web agents. Experimental results show that Ares reduces reasoning token usage by up to 52.7% compared to fixed high-effort reasoning, while introducing minimal degradation in task success rates.

Problem

• reasoning effort selection
• LLM agents
• inference cost
• dynamic reasoning
• multi-step tasks

Innovation

• dynamic reasoning effort
• lightweight router
• LLM agents
• cost-efficient inference
• step-level adaptation