Advancing sub-field artificial intelligence for open source1.8T data open-sourced.

Sleeping AI brings together European researchers and hackers to build targeted datasets, efficient architectures, and frontier models where resource investment creates maximal impact.

Directions

Scaling Laws
Audio & Music
TinyML
Efficient Architectures
Open Datasets
Multimodality

Widespread Impact

Inspired by Aaron Swartz and the early web's open ethos, we channel that radical passion into sub-field AI research. We focus on publishing long-term, maximum-impact open-source projects rather than transient, hype-driven work.

Governance

We operate as a decentralized collective—a hive mind spanning across Europe while remaining bound to no single institution. No individual holds unilateral authority; projects are planned, debated, and executed through autonomous task groups.

Index