GTM: Simulating the World of Tools for AI Agents

2025-12-04

AI Summary

To address the high computational cost, low efficiency, and substantial maintenance overhead associated with integrating external tools into LLM-based agents, this paper proposes the Generalist Tool Model (GTM)—a lightweight (1.5B-parameter) sequence model. GTM employs Context-Aware Response Generation (CARG) and a prompt-level data synthesis pipeline to faithfully emulate over 20,000 cross-domain tools. Crucially, GTM enables agent training without invoking real tools, achieving— for the first time—zero-shot cross-domain tool generalization and strong contextual consistency. Experiments demonstrate that GTM’s inference speed exceeds that of real tools by several orders of magnitude, while maintaining comparable output accuracy. Moreover, in reinforcement learning settings, GTM significantly enhances training scalability and deployment flexibility.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Multiagent Systems: Modeling other Agents
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions

Abstract

The integration of external tools is pivotal for empowering Large Language Model (LLM) agents with real-world capabilities. However, training these agents through direct, continuous interaction with diverse tools is often prohibitively expensive, slow, and introduces additional development and maintenance overhead. To address this challenge, we introduce the Generalist Tool Model (GTM), a 1.5-billion-parameter model that learns to act as a universal tool simulator. With only prompt-level configuration, GTM accesses tool functionalities along with input arguments and generates outputs that faithfully mimic real tool execution, providing a fast and cost-effective solution that eliminates development overhead. To build GTM, we propose the Context-Aware Response Generation (CARG) pipeline, which synthesizes comprehensive training data covering over 20,000 tools across 300 domains including physics, medicine, robotics, and finance. Through this pipeline, GTM learns to produce not only syntactically correct outputs but also logically coherent and contextually appropriate responses. Experiments demonstrate that GTM produces high-quality outputs with strong consistency and reliability. Besides when used in real reinforcement learning scenarios for agent training, GTM exhibits significantly faster simulation speed compared to real tools while maintaining comparable output quality, along with remarkable generalization and domain adaptability. Our results establish GTM as a foundational component for developing future AI agents, enabling efficient and scalable training of tool-augmented systems.

Problem

• Simulates diverse tools to reduce AI agent training costs
• Generates realistic tool outputs without direct tool integration
• Enables scalable training across multiple domains efficiently

Innovation

• GTM simulates diverse tools using 1.5B parameters
• CARG pipeline synthesizes training data across 300 domains
• GTM enables fast, cost-effective agent training via simulation