
Nonprofit · Hamburg, Germany
LAIONincl. OpenAssistant
German non-profit that releases large open datasets, tools and models for AI research, best known for LAION-400M, LAION-5B and OpenCLIP.
Hamburg, Germany
- Founded
- 2021
- Type
- Nonprofit
- Based in
- Hamburg, Germany
- Key papers
- 4
- Featured projects
- 7
About
LAION, the Large-scale Artificial Intelligence Open Network, is a non-profit research network based in Germany that makes machine learning resources available to the public. It is supported by donations and public research grants. Its volunteers include Christoph Schuhmann, Jenia Jitsev and Romain Beaumont.
In 2021 it released LAION-400M, an image-text dataset recreating the kind of data used for CLIP, followed by LAION-5B in 2022, the largest openly available image-text dataset at the time. Stable Diffusion was trained on it, and LAION members helped train OpenCLIP models on it. In 2023 it co-organised the OpenAssistant crowd-sourced conversation project.
After Stanford researchers reported suspected abuse material links in LAION-5B in December 2023, LAION withdrew its datasets and in August 2024 released a cleaned Re-LAION-5B.
Mission and ambitionsLAION states that it makes large-scale machine learning resources, including datasets, tools and open models, available to the public for research and education.
Known for
- 01
LAION-5B image-text dataset
- 02
OpenCLIP models
- 03
OpenAssistant
Milestones
- 2021
LAION-400M
Released 400 million image-text pairs filtered with CLIP.
- 2022
LAION-5B
Released over 5 billion image-text pairs, used to train Stable Diffusion.
- 2022
Reproducible scaling laws
Published OpenCLIP scaling-law research using LAION-2B.
- 2023
OpenAssistant
Released an open chat assistant and dataset built by volunteers.
- 2023
Dataset takedown
Removed LAION-5B and LAION-400M temporarily after a report of illegal content.
- 2024
Re-LAION-5B
Released a cleaned version of the dataset.
Open source
Chat-based assistant project; most-starred repo.
Contrastive Language-Audio Pretraining.
Evaluation for CLIP-like models.
Linear aesthetic estimator on CLIP embeddings.
- clap-htsat-fusedmodel
CLAP audio-text model.
OpenCLIP ViT-B/32 trained on LAION-2B.
OpenCLIP ViT-L/14 trained on LAION-2B.
Research papers
Notable releases
Openness
Datasets, models and code are open with mixed licences.
People
- Christoph SchuhmannCo-founder
- Jenia JitsevCo-founder, Jülich Supercomputing Centre
- Romain BeaumontCore contributor, OpenCLIP model training
- Richard VencuCo-founder
- Mehdi ChertiCo-founder
Listed from the lab's own pages and public profiles.
Contact and community
Contact details and links are on the lab's website (https://laion.ai); code questions go through issues on their GitHub organisation.
Public Discord linked from the lab's own website for community discussion and announcements.
Sources
Compiled from the lab's public pages, Hugging Face and GitHub. Last checked 2026-11-17. Something wrong or out of date? Tell us on Discord.



