# Iterative Tool Usage Exploration for Multimodal Agents via Step-wise Preference Tuning

arXiv: https://arxiv.org/abs/2504.21561 · HTML: https://mlsanity.com/paper/2504.21561/

## Key facts

- Cited by 19 papers in the map (data through 14 Oct 2026).
- Placed in Embodied and 3D vision-language models, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 56 of its 71 references that are in the map.
- First posted on arXiv 30 Apr 2025.

## Placed in

- Multimodal LLMs / vision-language models / [Embodied and 3D vision-language models](https://mlsanity.com/mllm/embodied-and-3d-vision-language-models/)
- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)
- LLM alignment / RLHF / preference optimisation / [Process reward models for reasoning](https://mlsanity.com/alignment/process-reward-models-for-reasoning/)

## Builds on (most cited first, 30 of 56)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023)
- [Qwen Technical Report](https://mlsanity.com/paper/2309.16609/) (2023)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Self-Instruct: Aligning Language Models with Self-Generated Instructions](https://mlsanity.com/paper/2212.10560/) (2022)
- [Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning](https://mlsanity.com/paper/2503.09516/) (2025)
- [GAIA: a benchmark for General AI Assistants](https://mlsanity.com/paper/2311.12983/) (2023)
- [MiniCPM-V: A GPT-4V Level MLLM on Your Phone](https://mlsanity.com/paper/2408.01800/) (2024)
- [How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suites](https://mlsanity.com/paper/2404.16821/) (2024)
- [ShareGPT4V: Improving Large Multi-Modal Models with Better Captions](https://mlsanity.com/paper/2311.12793/) (2023)
- [RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback](https://mlsanity.com/paper/2309.00267/) (2023)
- [ViperGPT: Visual Inference via Python Execution for Reasoning](https://mlsanity.com/paper/2303.08128/) (2023)
- [SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agents](https://mlsanity.com/paper/2401.10935/) (2024)
- [Visual Programming: Compositional visual reasoning without training](https://mlsanity.com/paper/2211.11559/) (2022)
- [Training Language Models to Self-Correct via Reinforcement Learning](https://mlsanity.com/paper/2409.12917/) (2024)
- [ReTool: Reinforcement Learning for Strategic Tool Use in LLMs](https://mlsanity.com/paper/2504.11536/) (2025)
- [AgentTuning: Enabling Generalized Agent Abilities for LLMs](https://mlsanity.com/paper/2310.12823/) (2023)
- [Step-DPO: Step-wise Preference Optimization for Long-chain Reasoning of LLMs](https://mlsanity.com/paper/2406.18629/) (2024)
- [WebRL: Training LLM Web Agents via Self-Evolving Online Curriculum Reinforcement Learning](https://mlsanity.com/paper/2411.02337/) (2024)
- [VideoAgent: A Memory-augmented Multimodal Agent for Video Understanding](https://mlsanity.com/paper/2403.11481/) (2024)
- [ArCHer: Training Language Model Agents via Hierarchical Multi-Turn RL](https://mlsanity.com/paper/2402.19446/) (2024)
- [DigiRL: Training In-The-Wild Device-Control Agents with Autonomous Reinforcement Learning](https://mlsanity.com/paper/2406.11896/) (2024)
- [Agent Q: Advanced Reasoning and Learning for Autonomous AI Agents](https://mlsanity.com/paper/2408.07199/) (2024)
- [APIGen: Automated Pipeline for Generating Verifiable and Diverse Function-Calling Datasets](https://mlsanity.com/paper/2406.18518/) (2024)
- [Trial and Error: Exploration-Based Trajectory Optimization for LLM Agents](https://mlsanity.com/paper/2403.02502/) (2024)
- [Vision-Language Models are Zero-Shot Reward Models for Reinforcement Learning](https://mlsanity.com/paper/2310.12921/) (2023)
- [Fine-Tuning Large Vision-Language Models as Decision-Making Agents via Reinforcement Learning](https://mlsanity.com/paper/2405.10292/) (2024)

## Built on by (most cited first, 19 of 19)

- [Tree Search for LLM Agent Reinforcement Learning](https://mlsanity.com/paper/2509.21240/) (2025)
- [Efficient Multi-turn RL for GUI Agents via Decoupled Training and Adaptive Data Curation](https://mlsanity.com/paper/2509.23866/) (2025)
- [TRACE: A Unified Rollout Budget Allocation Framework for Efficient Agentic Reinforcement Learning](https://mlsanity.com/paper/2606.11119/) (2026)
- [OmniRAG-Agent: Agentic Omnimodal Reasoning for Low-Resource Long Audio-Video Question Answering](https://mlsanity.com/paper/2602.03707/) (2026)
- [GUIDE: Resolving Domain Bias in GUI Agents through Real-Time Web Video Retrieval and Plug-and-Play Annotation](https://mlsanity.com/paper/2603.26266/) (2026)
- [Healthcare AI GYM for Medical Agents](https://mlsanity.com/paper/2605.02943/) (2026)
- [APPO: Agentic Procedural Policy Optimization](https://mlsanity.com/paper/2606.12384/) (2026)
- [Aligning Perception, Reasoning, Modeling and Interaction: A Survey on Physical AI](https://mlsanity.com/paper/2510.04978/) (2025)
- [Long-Horizon Visual Imitation Learning via Plan and Code Reflection](https://mlsanity.com/paper/2509.05368/) (2025)
- [Model Context Protocol for Vision Systems: Audit, Security, and Protocol Extensions](https://mlsanity.com/paper/2509.22814/) (2025)
- [Loong: A Human-Like Long Document Translation Agent with Observe-and-Act Adaptive Context Selection](https://mlsanity.com/paper/2605.30274/) (2026)
- [Allen: Rethinking MAS Design through Step-Level Policy Autonomy](https://mlsanity.com/paper/2508.11294/) (2025)
- [VTOS: Learning to Orchestrate Vision Tools by Co-Searching Solutions and Observers](https://mlsanity.com/paper/2606.20728/) (2026)
- [ReGRPO: Reflection-Augmented Policy Optimization for Tool-Using Agents](https://mlsanity.com/paper/2606.31392/) (2026)
- [WMLLM: Self-Evolving Optimization Agents via Predict-Then-Act World Modeling](https://mlsanity.com/paper/2609.01608/) (2026)
- [UnifiedPlayers: Enhance Tool-Integrated Reasoning in Agentic Reinforcement Learning](https://mlsanity.com/paper/2609.20089/) (2026)
- [Success Leaves Detours: Learning Executable Walkthroughs for Long-Horizon Agents](https://mlsanity.com/paper/2609.22120/) (2026)
- [V-Gym: Enhancing Agentic Visual Reasoning via Skill-Data Co-Evolution](https://mlsanity.com/paper/2609.34682/) (2026)
- [RSure-Agent: Reliable Use of Tool Observations for Remote Sensing Agents](https://mlsanity.com/paper/2610.04836/) (2026)

## Abstract (arXiv)

Multimodal agents, which integrate a controller e.g., a vision language model) with external tools, have demonstrated remarkable capabilities in tackling complex multimodal tasks. Existing approaches for training these agents, both supervised fine-tuning and reinforcement learning, depend on extensive human-annotated task-answer pairs and tool trajectories. However, for complex multimodal tasks, such annotations are prohibitively expensive or impractical to obtain. In this paper, we propose an iterative tool usage exploration method for multimodal agents without any pre-collected data, namely SPORT, via step-wise preference optimization to refine the trajectories of tool usage. Our method enables multimodal agents to autonomously discover effective tool usage strategies through self-exploration and optimization, eliminating the bottleneck of human annotation. SPORT has four iterative components: task synthesis, step sampling, step verification, and preference tuning. We first synthesize multimodal tasks using language models. Then, we introduce a novel trajectory exploration scheme, where step sampling and step verification are executed alternately to solve synthesized tasks. In step sampling, the agent tries different tools and obtains corresponding results. In step verification, we employ a verifier to provide AI feedback to construct step-wise preference data. The data is subsequently used to update the controller for tool usage through preference tuning, producing a SPORT agent. By interacting with real environments, the SPORT agent gradually evolves into a more refined and capable system. Evaluation in the GTA and GAIA benchmarks shows that the SPORT agent achieves 6.41% and 3.64% improvements, underscoring the generalization and effectiveness introduced by our method. The project page is https://SPORT-Agents.github.io.