How Good LLMs Are at Answering Bangla Medical Visual Questions? Dataset and Benchmarking

TL;DR

This paper introduces BanglaMedVQA, a new dataset for Bangla medical visual question answering, and evaluates current models, revealing significant performance gaps in low-resource language medical reasoning.

Contribution

The creation of BanglaMedVQA dataset and comprehensive benchmarking of foundation models on Bangla medical visual questions.

Findings

Current models perform poorly on Bangla MedVQA, especially on complex diagnostic questions.

Top models like GPT-4.1 mini still fail to accurately answer specialized medical questions.

Open-source models like Gemma-3 sometimes outperform top models in general categories.

Abstract

Recent advancements in Large Language Models (LLMs) and Large Vision Language Models (LVLMs) have enabled general-purpose systems to demonstrate promising capabilities in complex reasoning tasks, including those in the medical domain. Medical Visual Question Answering (MedVQA) has particularly benefited from these developments. However, despite Bangla being one of the most widely spoken languages globally, there exists no established MedVQA benchmark for it. To address this gap, we introduce BanglaMedVQA, a dataset comprising clinically validated image-question-answer pairs, along with a comprehensive evaluation of current foundation models on this resource. Consistent with prior findings that report low performance of current models on English MedVQA benchmarks, our analysis reveals that Bangla performance is substantially lower, reflecting the challenges inherent to low-resource languages. Even top-performing models such as Gemini and GPT-4.1 mini fail to accurately answer specialized diagnostic questions, indicating severe limitations in fine-grained medical reasoning. Although certain open-source models, such as Gemma-3, occasionally outperform these models in general categories, they too struggle with clinically complex questions, underscoring the urgent need for top-notch evaluation method.