LAION (Large-scale Artificial Intelligence Open Network) is a German nonprofit that builds and releases openly licensed datasets, tools, and models for machine learning research. It was founded in 2021 by Christoph Schuhmann and collaborators who met on a Discord server dedicated to AI, frustrated that the datasets behind breakthroughs like OpenAI's CLIP were not publicly available.
Rather than sell a product, LAION curates web-scraped image-caption pairs and other multimodal data at scale; its LAION-5B release in 2022 contained more than 5 billion image-text pairs and became the training foundation for Stable Diffusion and other open image generators. The organization also releases open-source models and tools, including work on the Open-Assistant chat project, through its GitHub organization.
LAION is built for AI researchers, open-source model developers, and academics who need training data at scale without commercial licensing restrictions. Everything LAION publishes is free to access and reuse under open licenses; there is no paid tier, subscription, or enterprise plan.
