What is DALL-E 3?

A text-to-image model is a machine learning system that takes a written description and produces a picture matching it. A hard part is faithfulness: OpenAI notes that such systems tend to ignore words or descriptions, which forces people to learn prompt engineering, the craft of wording requests so the system follows them. DALL-E 3 is an image-generation model from OpenAI, which the maker describes as built to generate images that adhere to the text it is given. OpenAI's announcement says it was built natively on ChatGPT, so that ChatGPT could act as a brainstorming partner and refiner of the prompt. OpenAI announced it in September 2023.

Written from primary sources and reference pages, and checked against them by an independent AI fact-checker. Drafted with AI assistance.