MGIE, which stands for MLLM-Guided Image Editing, is a model developed by Apple and the University of California, Santa Barbara, that leverages multimodal large language models (MLLMs) to enable image editing based on natural language instructions.