- Release date
- File format
- Parameter count
- Quantisation
- File size
- Supported operating systems
- Minimum memory
- Licence
- SHA-256 checksum
Introducing OpenATG
AI for the
greater good.
Useful intelligence, built to stay closer to people and the work that matters to them.
Compact by design.
Private by default.
Human at the centre.
Currently in development
Three systems.
One direction.
A compact foundation, a verifiable safety layer and a human-led application platform—designed as one responsible technology family.
ATG Base 32M
Small model. Serious intent.
A compact language model designed for everyday office work—from drafting and summarising to translation—without depending on the cloud.
See download plans- 32M
- parameter class
- Local
- designed to run close to the work
- Offline-first
- without depending on the cloud
- Built from first principles
- an original compact foundation
Not every useful model needs a data centre.
ATG Guard
Trust should be visible.
A privacy and accountability layer designed to make sensitive workflows easier to validate, trace and explain.
- Privacy validation
- Pseudonymised records
- Evidence hashes
- Replay-safe detection
Proof that follows the work.
ATG Apply
One profile. Every opportunity.
A human-led career system where candidates build one truthful profile, operators prepare tailored applications and nothing moves forward without the candidate's control.
- Human expertise
- Smart matching
- Secure applications
Downloads
Model downloads will live here.
ATG Base models are still being developed and validated. Verified releases, documentation and checksums will be published here when they are ready.
Why OpenATG exists
Technology should meet people where they are.
We build practical intelligence for ordinary computers, real institutions and human decisions. Less distance. More control. Better outcomes.
Contact
Let's build something useful.
For collaborations, research and conversations about the OpenATG family.
adrian.veluppillai@atgconcordia.com