COLT: Lightweight Multi-LLM Collaboration through Shared MCTS Reasoning for Model Compilation

2026-02-02

AI Summary

This work proposes COLT, a lightweight multi-LLM collaborative framework designed to reduce the high inference cost of large language models (LLMs) in compiler optimization while mitigating the reliability limitations of smaller models. COLT leverages a shared Monte Carlo Tree Search (MCTS) as a coordination backbone to enable cross-model inference reuse and value propagation. It incorporates model-aware tree policies and a curriculum-based adjustment mechanism to jointly select actions—comprising both optimization transformations and the next participating model. Compared to single-model approaches, COLT achieves comparable or superior performance with significantly lower computational overhead, all while avoiding the reliance on complex external components typical of conventional multi-agent systems.

Technology Category

• Multiagent Systems: Coordination and Collaboration
• Search and Optimization: Learning to Search
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Model serving costs dominate AI systems, making compiler optimization essential for scalable deployment. Recent works show that a large language model (LLM) can guide compiler search by reasoning over program structure and optimization history. However, using a single large model throughout the search is expensive, while smaller models are less reliable when used alone. Thus, this paper seeks to answer whether multi-LLM collaborative reasoning relying primarily on small LLMs can match or exceed the performance of a single large model. As such, we propose a lightweight collaborative multi-LLM framework, dubbed COLT, for compiler optimization that enables coordinated reasoning across multiple models within a single Monte Carlo tree search (MCTS) process. A key contribution is the use of a single shared MCTS tree as the collaboration substrate across LLMs, enabling the reuse of transformation prefixes and cross-model value propagation. Hence, we circumvent both heavy internal reasoning mechanisms and conventional agentic machinery that relies on external planners, multiple concurrent LLMs, databases, external memory/versioning of intermediate results, and controllers by simply endogenizing model selection within the lightweight MCTS optimization loop. Every iteration, the acting LLM proposes a joint action: (compiler transformation, model to be queried next). We also introduce a model-aware tree policy that biases search toward smaller models while preserving exploration, and a course-alteration mechanism that escalates to the largest model when the search exhibits persistent regressions attributable to smaller models.

Problem

• multi-LLM collaboration
• compiler optimization
• model serving cost
• lightweight reasoning
• shared MCTS

Innovation

• multi-LLM collaboration
• shared MCTS
• compiler optimization
• lightweight reasoning
• model-aware tree policy