GPT-4o

GPT-4o (“o” for “omni”) is a step towards much more natural human-computer interaction. It accepts as input any combination of text, audio, and image, and generates any combination of text, audio, and image outputs.