We’re announcing GPT-4o, our new flagship model that can reason across audio, vision, and text in real time.
Make a song about anything
Consistent Self-Attention for Long-Range Image and Video Generation
