# Kokoro-82M — card for AI agents

82M-parameter open-weight TTS model under Apache-2.0 with fixed voice packs.

Category: Voiceover (text-to-speech) · maker hexgrad · checked 2026-10-10

## Price
- Price: Free
- Free tier: Self-hosted; compute only.

## License
Apache-2.0 — https://huggingface.co/hexgrad/Kokoro-82M

- commercial: **yes** (Commercial use OK)
  > This is an Apache-licensed model, and Kokoro has been deployed in numerous projects and commercial APIs.
  (https://huggingface.co/hexgrad/Kokoro-82M)
- attribution: **no** (No credit needed)
- in app: **yes** (Can ship inside your app)
  > This is an Apache-licensed model, and Kokoro has been deployed in numerous projects and commercial APIs.
  (https://huggingface.co/hexgrad/Kokoro-82M)
- youtube: **yes** (OK in monetized YouTube videos)

## Rules to follow
- Keep the Apache-2.0 license notice when you redistribute the model.

## How an agent gets files
- pip install kokoro soundfile (also needs espeak-ng); load KPipeline(lang_code='a') and call it with text and voice='af_heart'. Weights on Hugging Face. (sign-in: none)
- Formats: wav
- Not allowed: Apache-2.0: keep the license notice when redistributing.
- Not allowed: Output is 24 kHz audio; v1.0 supports 8 languages and 54 voices.

## How to connect
- Package (npm/PyPI)
- API docs: https://github.com/hexgrad/kokoro
- npm: `npm install kokoro-js`
- PyPI: `pip install kokoro`
- Repository: https://github.com/hexgrad/kokoro
- Website: https://huggingface.co/hexgrad/Kokoro-82M

## Notes
No voice cloning. Voices cover English, Japanese, Mandarin, Spanish, French, Hindi and more; no Russian.

## Numbers (AgentGid data)
- Gid Score: 43.9
- GitHub stars: 9,547
- Downloads last week: 166,789
- Used in 2 public projects set up for AI coding agents

## Lines to paste into your prompt or AGENTS.md

```
- Kokoro-82M (Voiceover (text-to-speech)). 82M-parameter open-weight TTS model under Apache-2.0 with fixed voice packs. Price: free; commercial use OK; no credit needed.
  Connect: npm kokoro-js; PyPI kokoro.
  Get files: pip install kokoro soundfile (also needs espeak-ng); load KPipeline(lang_code='a') and call it with text and voice='af_heart'. Weights on Hugging Face (wav).
  Not allowed: Apache-2.0: keep the license notice when redistributing.
  Not allowed: Output is 24 kHz audio; v1.0 supports 8 languages and 54 voices.
  Rule: Keep the Apache-2.0 license notice when you redistribute the model.
  License: https://huggingface.co/hexgrad/Kokoro-82M (checked 2026-10-10; details: https://agentgid.com/assets/kokoro/)
```

Not verified: Last push to the GitHub repo was 2025-08-06.

## Links
- AgentGid page: https://agentgid.com/assets/kokoro/
- Is it free for commercial use? https://agentgid.com/assets/kokoro/commercial-use/
- Source: https://huggingface.co/hexgrad/Kokoro-82M
- Source: https://huggingface.co/hexgrad/Kokoro-82M/blob/main/VOICES.md
- Source: https://github.com/hexgrad/kokoro

---
AgentGid's catalog is curated: every entry was checked against the maker's official pages (registry, repository, license and pricing pages) on the date shown; a license value shows the quote it stands on wherever AgentGid has one. Usage numbers come from AgentGid's own scan of public repositories set up for AI coding agents. Claims by makers are reported, not tested by AgentGid.