DiffusionGPT is a unified text-to-image generation system that leverages Large Language Models (LLMs) to handle diverse input types, including prompt-based, instruction-based, inspiration-based, and hypothesis-based prompts. It integrates domain-expert models to generate high-quality images, using a Tree-of-Thought (ToT) structure to guide model selection based on input prompts. The system also incorporates Advantage Databases enriched with human feedback to align model selection with human preferences. DiffusionGPT addresses the limitations of existing text-to-image systems by providing a flexible and efficient framework that can adapt to various domains and input types. It outperforms traditional stable diffusion models in terms of image quality and aesthetic scores, demonstrating significant improvements in realism and detail. The system is training-free and can be easily integrated as a plug-and-play solution, offering a versatile and effective pathway for community development in image generation. Through extensive experiments and comparisons, DiffusionGPT has shown its effectiveness in generating high-quality images across various domains and prompt types. The system's ability to parse diverse prompts and select the most suitable model ensures exceptional performance and user satisfaction. DiffusionGPT's contributions include a new insight into using LLMs for text-to-image generation, an all-in-one system that integrates multiple diffusion models, and a training-free approach that enhances efficiency and flexibility. The system's use of ToT and human feedback improves accuracy and enables a more flexible process for aggregating expert models. Overall, DiffusionGPT represents a significant advancement in text-to-image generation, offering a robust and effective solution for generating high-quality images across diverse domains.DiffusionGPT is a unified text-to-image generation system that leverages Large Language Models (LLMs) to handle diverse input types, including prompt-based, instruction-based, inspiration-based, and hypothesis-based prompts. It integrates domain-expert models to generate high-quality images, using a Tree-of-Thought (ToT) structure to guide model selection based on input prompts. The system also incorporates Advantage Databases enriched with human feedback to align model selection with human preferences. DiffusionGPT addresses the limitations of existing text-to-image systems by providing a flexible and efficient framework that can adapt to various domains and input types. It outperforms traditional stable diffusion models in terms of image quality and aesthetic scores, demonstrating significant improvements in realism and detail. The system is training-free and can be easily integrated as a plug-and-play solution, offering a versatile and effective pathway for community development in image generation. Through extensive experiments and comparisons, DiffusionGPT has shown its effectiveness in generating high-quality images across various domains and prompt types. The system's ability to parse diverse prompts and select the most suitable model ensures exceptional performance and user satisfaction. DiffusionGPT's contributions include a new insight into using LLMs for text-to-image generation, an all-in-one system that integrates multiple diffusion models, and a training-free approach that enhances efficiency and flexibility. The system's use of ToT and human feedback improves accuracy and enables a more flexible process for aggregating expert models. Overall, DiffusionGPT represents a significant advancement in text-to-image generation, offering a robust and effective solution for generating high-quality images across diverse domains.