NML / Products
Products made by us
SOTA avatar models for offline generation and live conversation.
Model index / 2026
NML is developing multiple model families, evaluation systems and products. This index will expand as programs leave private research.
Atlas
Our first production model for enterprise digital humans.
Visit Atlas ↗Zankai
Our second-generation model, including anime and stylized characters.
In developmentAvatarBench
Benchmarks identity, motion, speech and visual fidelity.
In developmentDemo
Try realtime digital humans directly in the browser.
Launch demo ↗01
NORTH Realtime Avatar
One image in. A speaking human presence out.
NORTH turns any face photograph and audio stream into photorealistic, lip-synced video. Our SOTA one-shot system requires no identity training or preset avatar library.
| Offline | Face + audio or script → MP4 |
|---|---|
| Realtime | Live video over WebRTC |
| Input | One image; zero fine-tuning |
| Stack | Bring any LLM and TTS |
| Scale | Approximately $1–7 / generated hour, deployment dependent |
02 — NORTH ENTERPRISERealtime avatar infrastructure for enterprise AI products.View Enterprise →