AFAI Open Models

Models built for public impact.

Build and release technologies designed around public needs—especially where access, affordability, openness, or public benefit are not adequately served by the market.

Explore models

Open model stack

Focused capability

Efficient base model

Domain training

Transparent evaluation

Self-hosted deployment

Why open models

The frontier should not only live behind expensive APIs. Capable, specialized models can run where schools, researchers, nonprofits, and public institutions already work.

Model releases

Explore our models

1 model

Open releaseEducation

Specialized language model

Teach-1.0

Frontier pedagogical intelligence, democratized.

A specialized 4B-class tutoring model designed for misconception diagnosis, scaffolding, and sustained teaching sessions.

4B class

Model size

51.8

Reported MainScore

Local

Deployment

25.08.26

Release date

Development principles

Open is a practice, not only a download.

Useful releases require clear documentation, reproducible evaluation, responsible deployment guidance, and an honest account of what remains unknown.

Useful by design

Models are developed around a clear public need and evaluated against the work they are intended to support.

Efficient to deploy

Smaller, specialized systems can make capable AI available where expensive infrastructure or continuous API access is unrealistic.

Open responsibly

Releases pair access with intended-use guidance, limitations, risk documentation, and practical safeguards.

Document the evidence

Model cards distinguish measured results, evaluation assumptions, known gaps, and claims that still need independent validation.

What a release can include

Built to inspect, test, and adapt.

Artifacts vary by model and are published only when their release is appropriate.

Model weights

Quantized variants

Model card

Evaluation harness

Training recipe

Deployment guide

Build with us

Research, evaluate, and deploy public-interest models.

Contribute research, compute, evaluation, domain expertise, or deployment experience.