Atlas: Orchestrating Heterogeneous Models and Tools for Multi-Domain Complex Reasoning

2026-01-07

arXiv.org

AI Summary

This work addresses the challenge of dynamic scheduling for complex cross-domain reasoning in environments involving multiple large language models and external tools. The authors propose ATLAS, a novel framework that uniquely integrates two complementary routing mechanisms: training-free clustering-based routing and reinforcement learning–driven multi-step routing. This dual-path approach enables adaptive coordination of heterogeneous models and tools without requiring additional training. Evaluated across 15 benchmarks, ATLAS outperforms closed-source models such as GPT-4o, achieving a 10.1% performance gain on in-distribution tasks and a 13.1% improvement on out-of-distribution tasks. Notably, it also demonstrates substantial gains in visual reasoning, significantly enhancing cross-domain and out-of-distribution reasoning capabilities.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

The integration of large language models (LLMs) with external tools has significantly expanded the capabilities of AI agents. However, as the diversity of both LLMs and tools increases, selecting the optimal model-tool combination becomes a high-dimensional optimization challenge. Existing approaches often rely on a single model or fixed tool-calling logic, failing to exploit the performance variations across heterogeneous model-tool pairs. In this paper, we present ATLAS (Adaptive Tool-LLM Alignment and Synergistic Invocation), a dual-path framework for dynamic tool usage in cross-domain complex reasoning. ATLAS operates via a dual-path approach: (1) \textbf{training-free cluster-based routing} that exploits empirical priors for domain-specific alignment, and (2) \textbf{RL-based multi-step routing} that explores autonomous trajectories for out-of-distribution generalization. Extensive experiments across 15 benchmarks demonstrate that our method outperforms closed-source models like GPT-4o, surpassing existing routing methods on both in-distribution (+10.1%) and out-of-distribution (+13.1%) tasks. Furthermore, our framework shows significant gains in visual reasoning by orchestrating specialized multi-modal tools.

Problem

• heterogeneous models
• tool integration
• complex reasoning
• model selection
• multi-domain

Innovation

• heterogeneous model orchestration
• tool-LLM alignment
• cluster-based routing
• reinforcement learning routing
• multi-domain reasoning