FFA-GPT is an automated pipeline for fundus fluorescein angiography (FFA) interpretation and question-answering. The system combines a multi-modal transformer for image-text alignment and a large language model (LLM) for interactive QA. It was trained on 654,343 FFA images with 9392 reports and evaluated using both automatic and manual methods. The system generated coherent and comprehensible free-text reports with a BERTScore of 0.70 and F1 scores ranging from 0.64 to 0.82 for detecting top-5 retinal conditions. Manual evaluation showed acceptable accuracy (68.3%, Kappa 0.746) and completeness (62.3%, Kappa 0.739) of the generated reports. The generated answers were evaluated manually, with the majority meeting ophthalmologists' criteria. The system demonstrates potential to enhance ophthalmic image interpretation and facilitate interactive communications during medical consultations. The study introduces an innovative framework that combines multi-modal transformers and LLMs, enhancing ophthalmic image interpretation and facilitating interactive communications during medical consultation. The system is designed to reduce the reliance on retinal specialists and improve the efficiency of medical reporting. The model was evaluated using language-based and classification-based metrics, and the results showed that the system can generate accurate and complete reports. The system also provides interactive QA, which is beneficial for patients and doctors. The study highlights the potential of large language models in addressing challenges in FFA interpretation and medical consultations. The system is expected to improve the efficiency of medical reporting and reduce the workload of ophthalmologists. The study also discusses the limitations of the system, including the potential for hallucination and the need for further improvements in the model's performance. The system is intended to serve as an auxiliary tool to assist ophthalmologists in improving their work efficiency and accuracy, rather than replace their judgment and decision-making. The study concludes that the FFA-GPT system shows promising potential to enhance the interpretation and reporting of ophthalmic images, offering an important reference for the development of other image-based AI systems.FFA-GPT is an automated pipeline for fundus fluorescein angiography (FFA) interpretation and question-answering. The system combines a multi-modal transformer for image-text alignment and a large language model (LLM) for interactive QA. It was trained on 654,343 FFA images with 9392 reports and evaluated using both automatic and manual methods. The system generated coherent and comprehensible free-text reports with a BERTScore of 0.70 and F1 scores ranging from 0.64 to 0.82 for detecting top-5 retinal conditions. Manual evaluation showed acceptable accuracy (68.3%, Kappa 0.746) and completeness (62.3%, Kappa 0.739) of the generated reports. The generated answers were evaluated manually, with the majority meeting ophthalmologists' criteria. The system demonstrates potential to enhance ophthalmic image interpretation and facilitate interactive communications during medical consultations. The study introduces an innovative framework that combines multi-modal transformers and LLMs, enhancing ophthalmic image interpretation and facilitating interactive communications during medical consultation. The system is designed to reduce the reliance on retinal specialists and improve the efficiency of medical reporting. The model was evaluated using language-based and classification-based metrics, and the results showed that the system can generate accurate and complete reports. The system also provides interactive QA, which is beneficial for patients and doctors. The study highlights the potential of large language models in addressing challenges in FFA interpretation and medical consultations. The system is expected to improve the efficiency of medical reporting and reduce the workload of ophthalmologists. The study also discusses the limitations of the system, including the potential for hallucination and the need for further improvements in the model's performance. The system is intended to serve as an auxiliary tool to assist ophthalmologists in improving their work efficiency and accuracy, rather than replace their judgment and decision-making. The study concludes that the FFA-GPT system shows promising potential to enhance the interpretation and reporting of ophthalmic images, offering an important reference for the development of other image-based AI systems.