The Visual Question Answering (VQA) is a challenging application of AI that combines computer vision and natural language processing to answer questions based on visual content. This paper integrates recent developments in VQA, primarily focusing on text-based systems of VQA that add OCR to interpret textual elements present in images. New improvements like the Look, Read, Reason, and Answer (LoRRA) framework and datasets like TextVQA and ST-VQA have helped elevate the model’s ability to read and reason through text-centric tasks across modalities. The key developments include multimodal attention mechanisms, advanced techniques for text detection, and approaches for dynamic answer generation; these allow VQA systems to reach an unprecedented level in real-world applications. This survey offers a comparative evaluation of cutting-edge models, emphasizing how recent OCR-based models and multimodal fusion methods enhance text reading, reasoning accuracy, and robustness in practical applications. The discussion covers developments in attention-based mechanisms and answer prediction techniques and shows how more recent models excel beyond previous techniques in dealing with intricate text structures and noisy environments. This paper also tests the efficiency of dataset-driven improvement in enhancing model universality to varied languages and domains. Such applications may include accessibility enhancement through assistive technologies for visually impaired users, streamlined document processing in finance and law, and personalized learning advancement in educational platforms. However, the paper highlights challenges that need to be addressed, such as diversified orientations, noisy environments, and biased datasets. The result highlights the indispensable contribution of adopting text-informed learning methodologies and context-grounded reasoning for bolstering VQA model precision. The paper has actionable insights into and presents the system’s future directions, such as improved model generalization, efficiency, and domain adaptability to domains like education, healthcare, and assistive technologies to make robust applications possible. By bridging textual and visual reasoning, VQA systems pave the way to comprehensive AI-driven solutions.