Diffusion model: definition

A diffusion model is the kind of AI that generates images, and increasingly video, by starting from noise and refining it step by step into a result that matches a prompt.

Unlike language models that predict text, diffusion models create images. They learn to reverse a noising process, so from random noise and a text prompt they can produce a coherent image, which is how tools like modern image and video generators work.

For a business they power product imagery, marketing visuals, and design drafts. As with text, the useful pattern is to automate the production and keep a human on the creative judgement.

Related terms

Multimodal AI: Multimodal AI is a model that can take in or produce more than one kind of data, such as text plus images, audio, or video, rather than text alone.
Inference: Inference is the act of running a trained model to get an output, and inference cost is what you pay each time the model answers, measured in tokens.
Large language model (LLM): A large language model is a neural network trained on huge amounts of text to predict the next token, which lets it write, summarise, classify, and answer questions across almost any topic.