GPT-4 is the latest model addition to OpenAI's deep learning efforts and is a significant milestone in scaling deep learning. GPT-4 is also the first of the GPT models that is a large multimodal model, meaning it accepts both image and text inputs and emits text outputs.