Evidence-backed European AI register
Kyutai
Reviewed identity, roles, model listings, pricing and evidence for Kyutai.
Dataset compiled 2026-08-27 · evidence over inference
Reviewed identity
- Headquarters scope
- EU headquarters
- Headquarters record
- Paris
- Legal entity
- non-profit research laboratory (fonds de dotation)
- Ownership
- non-profit
- Founded
- 2023
- Confidence
- high
- Last verified
- 2026-08-22
Evidence boundaries
Organisation listings identify a reviewed canonical record. Model listings show an association in the source register; they do not by themselves establish model creation, current API availability, hosting location or infrastructure ownership.
Model listings
- Type
- text
- Parameters
- 7B (audio-native full-duplex speech dialogue)
- Context
- unknown
- Languages
- English (Moshi), French variants
- Licence
- CC-BY-4.0
- Release
- 2024-09
- Type
- asr
- Parameters
- neural audio codec
- Context
- unknown
- Languages
- n/a
- Licence
- CC-BY-4.0
- Release
- 2024-09
- Type
- text
- Parameters
- 2B / 6B
- Context
- unknown
- Languages
- multilingual — all 24 EU languages targeted
- Licence
- CC-BY-4.0
- Release
- 2025-01 (Helium 1)
- Hibiki / Hibiki-Zero
- Type
- text
- Parameters
- unknown
- Context
- unknown
- Languages
- FR↔EN speech translation
- Licence
- CC-BY-4.0
- Release
- 2025-02
- Type
- asr
- Parameters
- 1B, 2.6B
- Context
- unknown
- Languages
- English, French
- Licence
- CC-BY-4.0
- Release
- 2025-06
- Type
- tts
- Parameters
- Pocket TTS ~100M (CPU-capable)
- Context
- unknown
- Languages
- multilingual
- Licence
- CC-BY-4.0
- Release
- 2025-2026
- Unmute
- Type
- text
- Parameters
- unknown
- Context
- unknown
- Languages
- n/a
- Licence
- open source
- Release
- 2025-05
- MoshiVis / CASA
- Type
- multimodal
- Parameters
- unknown
- Context
- unknown
- Languages
- English
- Licence
- CC-BY-4.0
- Release
- 2025-03
- MuScriptor
- Type
- text
- Parameters
- unknown
- Context
- unknown
- Languages
- n/a (music)
- Licence
- unknown
- Release
- 2026
- ARC4/ARC8 Encoders (Llama, Mistral, multi)
- Type
- embedding
- Parameters
- 3B
- Context
- unknown
- Languages
- unknown
- Licence
- unknown
- Release
- 2026
- moshika-rl-seamless / personaplex-rl-seamless
- Type
- text
- Parameters
- 8B
- Context
- unknown
- Languages
- unknown
- Licence
- unknown
- Release
- 2026
Profile note
Europe's largest privately-endowed open-science AI lab and the world leader in open real-time speech-native models (Moshi, Mimi, Unmute, Kyutai STT/TTS). Everything ships open weights with no commercial gate — a deliberate counterweight to closed US voice stacks.
Record sources
- https://kyutai.org/
- https://huggingface.co/kyutai
- https://huggingface.co/kyutai/moshiko-pytorch-bf16
- https://www.iliad.fr/en/actualites/article/launch-of-kyutai-europe-s-first-independent-research-lab-dedicated…
- https://fortune.com/2023/11/17/ai-billionaires-xavier-niel-rodolphe-saade-eric-schmidt-france-lab/