the AI bench
VERIFIED JULY 2026
All models

MODEL · SUPERTONE · LIGHTWEIGHT (ON-DEVICE)

Supertonic 3

A lightweight on-device TTS that runs entirely locally through ONNX Runtime — no cloud call for synthesis. Supertonic 3 expands the open-weight release from 5 to 31 languages, improves reading stability, and cuts repeat/skip failures. `pip install supertonic` and synthesize immediately with selectable voice styles.

License: OpenRAIL (responsible-use clauses) · Context: n/a · Released: May 6, 2026

The decision in five lines

The call
Skip for local
Best for
Local evaluation and family reference
Runs on
No planner hardware fits at default quant — see model card
Watch out
Voice cloning from a reference clip or top-tier expressive quality — Chatterbox, VoxCPM2, or MOSS-TTS-v1.5 are the picks there.
Evidence
Editorial · last verified July 2026

Lightweight (on-device)
PARAMETERS
TTS
TYPE
CONTEXT
CPU / tiny — runs on-device via ONNX Runtime
VRAM AT Q4

Where we recommend this

This model isn’t currently in an active planner slot. See the runner notes below if you’re running it anyway.

The call

A lightweight on-device TTS that runs entirely locally through ONNX Runtime — no cloud call for synthesis. Supertonic 3 expands the open-weight release from 5 to 31 languages, improves reading stability, and cuts repeat/skip failures. `pip install supertonic` and synthesize immediately with selectable voice styles.

When not to use: Voice cloning from a reference clip or top-tier expressive quality — Chatterbox, VoxCPM2, or MOSS-TTS-v1.5 are the picks there. Supertonic's edge is speed + on-device footprint + broad language coverage, not cloning fidelity.

Runner notes

Python SDK (`pip install supertonic`) on ONNX Runtime — CPU-friendly, no GPU required. OpenRAIL license carries responsible-use terms; check them for commercial deploys. 31 languages.

License
OpenRAIL (responsible-use clauses)
Released
May 6, 2026
Maker
Supertone

Next step

Find-by-model — see what hardware runs this