Model · hexgrad · Text-To-Speech
Kokoro-82M
A tiny open text-to-speech model (82 million parameters) with 54 preset voices in 8 languages. Quality close to much larger models at a fraction of the cost — the most-downloaded TTS model on Hugging Face.
- Modality
- Text-To-Speech
- License
- Apache 2.0 (Open)
- Parameter size
- 82M
- Released
- January 27, 2025
- Last verified
- October 28, 2026
- Runs locally
- Yes
What you need to run it
- Mac
- Yes. Any Apple Silicon Mac (M1 or newer) with at least 8 GB of memory. Newer chips are faster, but memory is what decides whether it runs.
- Windows / Linux PC
- NVIDIA works best. A graphics card with at least 4 GB of video memory.
- No graphics card
- Works. Runs on a normal computer's processor, just slower.
Tiny enough to run on any laptop or CPU; the README documents Apple Silicon GPU use and MLX builds exist.
Numbers are for: official 82M weights. Our estimate from the model’s size — the maker publishes no figure.
Strengths
- Apache 2.0 weights, trained only on permissive or non-copyrighted audio
- Very small — 82M parameters, far lighter than any other open TTS here
- 54 preset voices across 8 languages
- Hosted copies are cheap: the card cites under $1 per million characters
- Runs on Apple Silicon GPUs with PyTorch's MPS fallback
Weaknesses
- Preset voices only — the card offers no voice-cloning mode
- Needs the espeak-ng system package for English fallback and some non-English languages
- No new model since v1.0 (2025-01-27); the GitHub repo was last pushed 2025-08
- Scam sites use the Kokoro name — only the hexgrad Hugging Face and GitHub pages are official
Try it
| Where | Type | Notes |
|---|---|---|
| Hugging Face | weights | Apache 2.0 |
| GitHub (kokoro package) | local | pip install kokoro; needs espeak-ng |
Official sources
- Model card model
- GitHub github
Change log
- — Added hardware requirements (Mac support, memory, NVIDIA). Source: https://github.com/hexgrad/kokoro + mlx-community/Kokoro-82M-bf16.
- — Initial entry. Added 2026-10-28 from the popularity sweep: the top text-to-speech model on Hugging Face by likes (7,292) and downloads (10.6M/month); 9.7k GitHub stars. v1.0 published 2025-01-27 per the model card's release table, following v0.19 (2024-12-25).
Esc