Intel / article

Moka v1

Source: million.dev
Author: aidenybai
Date: Published Jul 28, 2026

Why it matters

It's a hard, verifiable demonstration of how far task-specific can go — a 127x smaller student retaining most of the teacher's playing strength in a 101 KB browser-loadable ONNX file — useful reference material for anyone sizing a small model for edge or in-browser .

Terms in this piece · Glossary

distillation — Training a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.
inference — Running a trained model to get answers — the phase where AI is actually used, as opposed to trained.

Read the source million.dev