Artificial intelligence-based large language models (LLMs) are increasingly used in ophthalmic care, offering potential benefits in patient information, disease diagnosis, symptom triage, and education. However, their performance varies depending on the task domain and the specific LLM used. Human experts demonstrated the highest accuracy in disease diagnosis (86%), while ChatGPT-4 performed well in ophthalmology examinations (75.9%), symptom triage (98%), and providing information (84.6%). LLMs showed superior performance in general ophthalmology but were less accurate in subspecialties. Despite their efficiency, LLMs have limitations, including the potential to generate fake responses, lack of image processing capability, and inability to analyze critical literature. They also raise ethical and copyright concerns. LLMs can assist in various applications, such as generating information, answering questions, and supporting clinical decision-making. However, they are not yet reliable for critical tasks like diagnosing eye diseases or providing medical advice. Their performance in ophthalmic board examinations was comparable to human experts, but they still lag behind in accuracy. LLMs can also be used for scientific writing, drug discovery, and gene discovery, but their use in these areas is still limited. They have the potential to improve healthcare by providing accurate and timely information, but their integration into clinical practice requires careful evaluation. The future of LLMs in ophthalmic care depends on addressing their limitations, such as improving accuracy, ensuring ethical use, and enhancing their ability to process images and understand complex medical concepts. While LLMs can serve as valuable tools for education and research, they should not replace human expertise in clinical decision-making. Ophthalmic care professionals should use LLMs cautiously, relying on human judgment for critical decisions. The development of domain-specific LLMs tailored for ophthalmology could enhance their effectiveness in clinical settings. Overall, LLMs have the potential to improve ophthalmic care, but their implementation requires careful consideration of their strengths and limitations.Artificial intelligence-based large language models (LLMs) are increasingly used in ophthalmic care, offering potential benefits in patient information, disease diagnosis, symptom triage, and education. However, their performance varies depending on the task domain and the specific LLM used. Human experts demonstrated the highest accuracy in disease diagnosis (86%), while ChatGPT-4 performed well in ophthalmology examinations (75.9%), symptom triage (98%), and providing information (84.6%). LLMs showed superior performance in general ophthalmology but were less accurate in subspecialties. Despite their efficiency, LLMs have limitations, including the potential to generate fake responses, lack of image processing capability, and inability to analyze critical literature. They also raise ethical and copyright concerns. LLMs can assist in various applications, such as generating information, answering questions, and supporting clinical decision-making. However, they are not yet reliable for critical tasks like diagnosing eye diseases or providing medical advice. Their performance in ophthalmic board examinations was comparable to human experts, but they still lag behind in accuracy. LLMs can also be used for scientific writing, drug discovery, and gene discovery, but their use in these areas is still limited. They have the potential to improve healthcare by providing accurate and timely information, but their integration into clinical practice requires careful evaluation. The future of LLMs in ophthalmic care depends on addressing their limitations, such as improving accuracy, ensuring ethical use, and enhancing their ability to process images and understand complex medical concepts. While LLMs can serve as valuable tools for education and research, they should not replace human expertise in clinical decision-making. Ophthalmic care professionals should use LLMs cautiously, relying on human judgment for critical decisions. The development of domain-specific LLMs tailored for ophthalmology could enhance their effectiveness in clinical settings. Overall, LLMs have the potential to improve ophthalmic care, but their implementation requires careful consideration of their strengths and limitations.