we are an applied AI lab focused on building tools and models that you want to actually work with. models that can point things out, switch registers, look around, be guided, and know when they don't know something.
coming soon
- Phosphophyllite, a photobashing program
- Pyranine, a model that works with text
- Beryl, a model that can look around
- Agnotologist, a model that knows when it doesn't know
- Kototama, a hands-on inference toolkit built on llama.cpp