CharTool: Tool-Integrated Visual Reasoning for Chart Understanding

TL;DR

CharTool enhances multimodal large language models' ability to understand and reason about charts by integrating external tools and training on diverse synthesized and real-world chart data.

Contribution

It introduces DuoChart data pipeline and CharTool framework, enabling tool-augmented reasoning for improved chart understanding in MLLMs.

Findings

CharTool-7B outperforms base model by +8.0% on CharXiv reasoning.

CharTool achieves +9.78% on ChartQAPro.

Method generalizes well to out-of-domain visual math benchmarks.

Abstract

Charts are ubiquitous in scientific and financial literature for presenting structured data. However, chart reasoning remains challenging for multimodal large language models (MLLMs) due to the lack of high-quality training data, as well as the need for fine-grained visual grounding and precise numerical computation. To address these challenges, we first propose DuoChart, a scalable dual-source data pipeline that combines synthesized charts with real-world charts to construct diverse, high-quality chart training data. We then introduce CharTool, which equips MLLMs with external tools, including image cropping for localized visual perception and code-based computation for accurate numerical reasoning. Through agentic reinforcement learning on DuoChart, CharTool learns tool-integrated reasoning grounded in chart content. Extensive experiments on six chart benchmarks show that our method consistently improves over strong MLLM baselines across model scales. Notably, CharTool-7B outperforms the base model by **+8.0%** on CharXiv (Reasoning) and **+9.78%** on ChartQAPro, while achieving competitive performance with substantially larger or proprietary models. Moreover, CharTool demonstrates positive generalization to out-of-domain visual math reasoning benchmarks.