Coqui

weights

open(XTTS-v2 .pth checkpoints on the Hub, ungated)

data

closed(no training-data statement on either card)

code

partial(inference plus a GPT-encoder fine-tuning recipe in the maintained Idiap fork

license

Coqui-Public-Model-License-1.0.0(XTTS-v2 and XTTS-v1 weights

XTTS-v2 downloads freely, but the Coqui Public Model License allows only non-commercial use of the model and its outputs. Nothing is published about the training data, and only a fine-tuning recipe for one component is available. That license permits no commercial use, so its tier is commercial_forbidden.

• https://huggingface.co/api/models/coqui/XTTS-v2 recorded 2026-09-27

  Hub record: "gated":false, tag "license:other", and model.pth in the file list.

• https://huggingface.co/coqui/XTTS-v2/raw/main/LICENSE.txt recorded 2026-09-27

  License file titled "# Coqui Public Model License 1.0.0": "This license allows only non-commercial use of a machine learning model and its outputs."; "Use of the model to train other models for commercial use is not a non-commercial purpose."

• https://huggingface.co/coqui/XTTS-v2/raw/main/README.md recorded 2026-09-27

  Card: "This model is licensed under [Coqui Public Model License](https://coqui.ai/cpml)."; "The [code-base](https://github.com/coqui-ai/TTS) supports inference and [fine-tuning]".

• https://raw.githubusercontent.com/idiap/coqui-ai-TTS/HEAD/docs/source/models/xtts.md recorded 2026-09-27

  Idiap fork docs: "Current implementation only supports inference and GPT encoder training."

• https://raw.githubusercontent.com/idiap/coqui-ai-TTS/HEAD/README.md recorded 2026-09-27

  Idiap fork README: "Fork of the [original, unmaintained repository](https://github.com/coqui-ai/TTS)."

Hugging Face downloads of the XTTS checkpoints; installs through the Coqui TTS packages fetch the same files.

• https://huggingface.co/api/models?author=coqui&limit=100 recorded 2026-09-27

  Hub listing of the coqui organization: XTTS-v2 with "downloads":6781520 and XTTS-v1 with "downloads":1568.

Still the most downloaded open voice-cloning model, but listeners in blind tests rank it near the bottom of the board, a step below Kokoro.

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

  Leaderboard entry "name":"XTTS v2" (creator Coqui) with "elo":916.4 and "openWeights":true; 91 models are ranked.