RAY AI
A 3.2-billion-parameter language model built in-house by the GHINIX team over roughly three years — usable from the…
- 3.2B parameters
- ~3 years in development
- CLI ready
3.2B
Parameters
~3 yrs
In development
In-house
Built
Why it matters
- 01
Built from scratch in-house
Model design, training pipeline and tuning were done by the GHINIX team…
- 02
Compact 3.2B parameters
Small enough to self-host on a single server, so sensitive data does…
- 03
Works in the terminal
RAY AI runs from a command-line interface today, for developers who live…
- 04
Custom deployments
We can adapt RAY AI to your documents and workflows as part…
About the model
RAY AI is our own large language model. It was designed, trained and refined by the GHINIX engineering team, not wrapped around someone else's API.
At 3.2B parameters it is compact enough to run on modest hardware, which makes private and on-premise deployment practical for Indian businesses that want to keep their data inside their own network.
RAY AI already runs in a CLI. We are now preparing a public command-line tool — India's own CLI built on an Indian model.
India's own CLI — coming soon
A public command-line tool powered by RAY AI. Leave your details and we will tell you when early access opens.
$ ray "summarise this log file" < server.logGet notified