DALL·E
A 12-billion parameter version of GPT-3 that generates images from text captions.
We’ve trained a neural network called DALL·E that creates images from text captions for a wide range of concepts expressible in natural language.
Official source
DALL·E: Creating images from textopenai.com · published