Craiyon, formerly known as DALL·E mini, is an AI model that can generate images from any text prompt. The model, developed by the craiyon team, including Boris Dayma and Pedro Cuenca, is trained using Google TRC and the larger version is known as DALL·E mega. The model relies on ads and donations to pay for servers, as it requires a lot of compute. Users are encouraged to be specific when creating images and to use keywords such as "illustration," "photorealistic," and "high definition" for better results. However, users should be aware that the model may reinforce or exacerbate societal biases, as it was trained on unfiltered data from the internet.