Kokoro-82M
An 82M-parameter Apache-2.0 TTS model that runs on CPU and topped the TTS Arena.
About
Kokoro-82M is an open-weight text-to-speech model from independent researcher hexgrad, with v1.0 released January 27, 2025. At just 82M parameters (StyleTTS 2 architecture with an iSTFTNet-derived decoder), it ships 54 voices across 8 languages at 24kHz, runs in real time on CPU — even decade-old desktop chips — and at launch topped the TTS Spaces Arena Elo ranking ahead of models over ten times its size.
The weights are Apache 2.0, so commercial use is fully permitted, and the model card warns that kokoro-branded websites are scams — the real weights live only on Hugging Face. It uses preset voices rather than voice cloning, and reviewers note it is weaker on dramatic emotional delivery than larger expressive TTS systems.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
- #1 in TTS Spaces Arena at v0.19 launch, ahead of XTTS v2 and others (project-reported)
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Viral community adoption; #1 in TTS Spaces Arena at launch; broad API hosting; Apache 2.0; Preset voices only, no voice cloning; weaker dramatic delivery than larger expressive TTS. License: Apache 2.0.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 7.0/10
- Openness · license + docs
- 10.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · Viral community adoption; #1 in TTS Spaces Arena at launch; broad API hosting
- · Apache 2.0
- · Preset voices only, no voice cloning; weaker dramatic delivery than larger expressive TTS