Beyond APIs: Probing the Limits of MLLMs in Physical Tool Use

• URL: http://arxiv.org/abs/2606.10803v1
• Date: Tue, 09 Jun 2026 12:49:11 GMT
• Title: Beyond APIs: Probing the Limits of MLLMs in Physical Tool Use
• Abstract summary: We introduce PhysTool-Bench, the first physical tool-use benchmark to evaluate MLLMs' ability to comprehend real-world scenarios, identify physical tools, and plan their use.<n>Across 13 leading MLLMs, even the strongest model (Gemini-3.1-Pro) identifies only 58.7% of tools in a scene and completes merely 21.0% of queries end-to-end.<n>Our analysis reveals a two-level deficit: MLLMs struggle to perceive tools in realistic scenes, and the much larger drop at the planning stage indicates a lack of functional commonsense.
• Score: 30.728333023657537
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Multimodal Large Language Models (MLLMs) excel at utilizing digital APIs and increasingly serve as the "brain" of embodied AI, instructing robots to interact with the physical world. In such embodied settings, a central capability is the use of physical tools, which underpins MLLMs' ability to assist humans in real-world tasks. Despite the importance, MLLMs' proficiency in physical tool use remains largely unexplored. To address this gap, we introduce PhysTool-Bench, the first physical tool-use benchmark designed to evaluate MLLMs' ability to comprehend real-world scenarios, identify physical tools, and plan their use. PhysTool-Bench comprises 2,510 queries over 2,678 real-world physical tools spanning diverse domains, including manufacturing, electrical work, agriculture, and healthcare. Concretely, models are evaluated along two primary dimensions: 1) recognizing all physical tools present in the scene, and 2) planning the tool selection and use sequence based on the instruction and visual context. Across 13 leading MLLMs, even the strongest model (Gemini-3.1-Pro) identifies only 58.7% of tools in a scene and completes merely 21.0% of queries end-to-end. Our analysis reveals a two-level deficit: MLLMs struggle to perceive tools in realistic scenes, and the much larger drop at the planning stage further indicates a lack of functional commonsense for mapping perceived tools onto task semantics, pinpointing a critical bottleneck for the development of practical embodied AI.

Related papers

• MetaToolAgent: Towards Generalizable Tool Usage in LLMs through Meta-Learning [16.060518943785514]
  We introduce a dataset spanning 7 domains, containing 155 tools and 9,377 question-answer pairs.<n>We also propose MetaToolAgent (MTA), a meta-learning approach designed to improve cross-tool generalization.
  arXiv  Detail & Related papers  (2026-01-19T02:53:31Z)
• PhysToolBench: Benchmarking Physical Tool Understanding for MLLMs [46.62285836344724]
  We present PhysToolBench, the first benchmark dedicated to evaluating the comprehension of physical tools by MLLMs.<n>Our benchmark is structured as a Visual Question Answering (VQA) dataset comprising over 1,000 image-text pairs.<n>It assesses capabilities across three distinct difficulty levels: (1) Tool Recognition: Requiring the recognition of a tool's primary function.<n> (2) Tool Understanding: Testing the ability to grasp the underlying principles of a tool's operation.
  arXiv  Detail & Related papers  (2025-10-10T16:10:45Z)
• MetaTool: Facilitating Large Language Models to Master Tools with Meta-task Augmentation [25.360660222418183]
  We present MetaTool, a novel tool learning methodology designed to generalize across any reusable toolset. By incorporating meta-task data into task-oriented training, our method significantly enhances the performance of open-source Large Language Models.
  arXiv  Detail & Related papers  (2024-07-15T10:15:41Z)
• GTA: A Benchmark for General Tool Agents [32.443456248222695]
  We design 229 real-world tasks and executable tool chains to evaluate mainstream large language models (LLMs) Our findings show that real-world user queries are challenging for existing LLMs, with GPT-4 completing less than 50% of the tasks and most LLMs achieving below 25%. This evaluation reveals the bottlenecks in the tool-use capabilities of current LLMs in real-world scenarios, which provides future direction for advancing general-purpose tool agents.
  arXiv  Detail & Related papers  (2024-07-11T17:50:09Z)
• Tool Learning in the Wild: Empowering Language Models as Automatic Tool Agents [56.822238860147024]
  Augmenting large language models with external tools has emerged as a promising approach to extend their utility.<n>Previous methods manually parse tool documentation and create in-context demonstrations, transforming tools into structured formats for LLMs to use in their step-by-step reasoning.<n>We propose AutoTools, a framework that enables LLMs to automate the tool-use workflow.
  arXiv  Detail & Related papers  (2024-05-26T11:40:58Z)
• What Are Tools Anyway? A Survey from the Language Model Perspective [67.18843218893416]
  Language models (LMs) are powerful yet mostly for text generation tasks. We provide a unified definition of tools as external programs used by LMs. We empirically study the efficiency of various tooling methods.
  arXiv  Detail & Related papers  (2024-03-18T17:20:07Z)
• Planning, Creation, Usage: Benchmarking LLMs for Comprehensive Tool Utilization in Real-World Complex Scenarios [93.68764280953624]
  UltraTool is a novel benchmark designed to improve and evaluate Large Language Models' ability in tool utilization. It emphasizes real-world complexities, demanding accurate, multi-step planning for effective problem-solving. A key feature of UltraTool is its independent evaluation of planning with natural language, which happens before tool usage.
  arXiv  Detail & Related papers  (2024-01-30T16:52:56Z)
• ToolEyes: Fine-Grained Evaluation for Tool Learning Capabilities of Large Language Models in Real-world Scenarios [49.33633818046644]
  We propose ToolEyes, a fine-grained system tailored for the evaluation of large language models' tool learning capabilities in authentic scenarios.<n>The system meticulously examines seven real-world scenarios, analyzing five dimensions crucial to LLMs in tool learning.<n>ToolEyes incorporates a tool library boasting approximately 600 tools, serving as an intermediary between LLMs and the physical world.
  arXiv  Detail & Related papers  (2024-01-01T12:49:36Z)
• Large Language Models as Tool Makers [85.00361145117293]
  We introduce a closed-loop framework, referred to as LLMs A s Tool Makers (LATM), where LLMs create their own reusable tools for problem-solving. Our approach consists of two phases: 1) tool making: an LLM acts as the tool maker that crafts tools for a set of tasks. 2) tool using: another LLM acts as the tool user, which applies the tool built by the tool maker for problem-solving.
  arXiv  Detail & Related papers  (2023-05-26T17:50:11Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.