hexgrad

weights

open(.pth checkpoint and voice packs, ungated)

data

described(a few hundred hours of permissively licensed audio, listed in part)

code

partial(inference library only)

license

Apache-2.0(OSI)

Kokoro's weights and inference library are Apache-2.0, an OSI-approved license. The card lists some of the openly licensed audio it trained on and sizes the rest at a few hundred hours, but the training set and training code are not published.

• https://huggingface.co/api/models/hexgrad/Kokoro-82M recorded 2026-09-27

  "license:apache-2.0" tag, "gated":false, with kokoro-v1_0.pth in the file list.

• https://huggingface.co/hexgrad/Kokoro-82M/raw/main/README.md recorded 2026-09-27

  "Audio licensed under Apache, MIT, etc"; "**Total Dataset Size:** A few hundred hours of audio".

• https://raw.githubusercontent.com/hexgrad/kokoro/HEAD/LICENSE recorded 2026-09-27

  LICENSE body is the Apache License, Version 2.0 text: "Apache License".

• https://raw.githubusercontent.com/hexgrad/kokoro/HEAD/README.md recorded 2026-09-27

  "An inference library for [Kokoro-82M](https://huggingface.co/hexgrad/Kokoro-82M)."

Hugging Face downloads of the single weights repository. The kokoro package is marked as not a separate channel because it loads this repository.

• https://huggingface.co/api/models/hexgrad/Kokoro-82M recorded 2026-09-27

  "downloads":11674701 for hexgrad/Kokoro-82M.

Kokoro is mid-table in blind listening tests, well behind the commercial leaders, but it is by far the smallest model near that position.

• https://artificialanalysis.ai/text-to-speech/leaderboard recorded 2026-09-27

  Leaderboard entry "name":"Kokoro 82M v1.0" with "elo":1065.14 and "openWeights":true; 91 models are ranked.