OpenThinkIMG: Learning to Think with Images via Visual Tool Reinforcement Learning

TL;DR

OpenThinkIMG introduces an open-source framework and a novel reinforcement learning method, V-ToolRL, enabling vision-language models to learn adaptive visual tool usage for complex reasoning tasks.

Contribution

It provides the first comprehensive infrastructure for tool-augmented LVLMs and proposes V-ToolRL, a reinforcement learning approach for dynamic tool invocation.

Findings

V-ToolRL significantly improves task success rates (+28.83 points) over supervised fine-tuning.

The RL agent outperforms baseline models like Taco and CogCom by +12.7 points.

The approach surpasses GPT-4.1 in accuracy by +8.68 points.

Abstract

While humans can flexibly leverage interactive visual cognition for complex problem-solving, enabling Large Vision-Language Models (LVLMs) to learn similarly adaptive behaviors with visual tools remains challenging. A significant hurdle is the current lack of standardized infrastructure, which hinders integrating diverse tools, generating rich interaction data, and training robust agents effectively. To address these gaps, we introduce OpenThinkIMG, the first open-source, comprehensive end-to-end framework for tool-augmented LVLMs. It features standardized vision tool interfaces, scalable trajectory generation for policy initialization, and a flexible training environment. Furthermore, considering supervised fine-tuning (SFT) on static demonstrations offers limited policy generalization for dynamic tool invocation, we propose a novel reinforcement learning (RL) framework V-ToolRL to train LVLMs to learn adaptive policies for invoking external vision tools. V-ToolRL enables LVLMs to autonomously discover optimal tool-usage strategies by directly optimizing for task success using feedback from tool interactions. We empirically validate V-ToolRL on challenging chart reasoning tasks. Our RL-trained agent, built upon a Qwen2-VL-2B, significantly outperforms its SFT-initialized counterpart (+28.83 points) and surpasses established supervised tool-learning baselines like Taco and CogCom by an average of +12.7 points. Notably, it also surpasses prominent closed-source models like GPT-4.1 by +8.68 accuracy points. We hope OpenThinkIMG can serve as a foundational framework for advancing dynamic, tool-augmented visual reasoning, helping the community develop AI agents that can genuinely "think with images".