Blogs
Browse RuVI Labs news and announcements
We are interested in building Language model from scratch & finetune models into its last limits
Building good models for people.
An independent research lab in India, working with collaborators around the world. We train language models from scratch.
RuVI Labs is a small research lab in India. We pretrain, finetune, and post-train language models, and we try to build things that work differently from the usual approach.
Pragya is our family of language models, trained end to end on open data.
| Model | Status | Notes |
|---|---|---|
| Pragya Bodh (flagship) | Planned | Open weights, open data |
| Pragya Vayu | Training | Built for low-resource hardware |
| Pragya mēdhā | Planned | Medium tier |
Open weights. Anyone can use and build on what we release.
Release what's strong. We ship models when they're good enough to stand up to real use.
Good models come from training them for a purpose, not from scale alone.
| Lab | RuVI Labs |
| Founder | Arush Kumar |
| Home | 🇮🇳 India |
| Family | Pragya |
| License | Open weights |
| Mode | Building quietly, shipping publicly |
Several groups have contributed compute, time, or support to this work.
Follow the organization on Hugging Face to get notified when models and datasets are released.
RUVI LABS
Small lab, good work.