About
A small research lab, studying how AI actually works.
Zellige AI is an independent research lab. We publish what we learn about language models and build our own from it.
Models
From research
to models
When our research leads somewhere useful, we turn it into models. Mosaic Pro and Mosaic Micro are our first two foundation models, both still in development; we plan to release open weights and offer chat access through a waitlist when they're ready. Tessera applies the same research to existing models, starting with reasoning in fewer tokens. The research comes first; the models follow from it.
Mosaic Pro
The larger model in the family.
In development
Mosaic Micro
The compact model in the family.
In development
Principles
How we work
Three principles shape what we do and what we don't.
- 01
Deliberately open
Research papers, open weights, occasionally code. We publish the experiments that fail alongside the ones that work; the systems that produce them are the lab's own.
- 02
Mechanistic rigor
We study how models actually work before we try to change them, and we measure every result against the model it replaces. No benchmark theatrics.
- 03
Experimental
We try approaches that haven't been widely studied. Sometimes they don't work. When they do, they tend to be the things that matter.
The lab
Team
We're a small team building in public. Team pages and public hires are coming. Reach out through our contact page if you want to collaborate.