Evidence-backed European AI register
Inria — ALMAnaCH team
Reviewed identity, roles, model listings, pricing and evidence for Inria — ALMAnaCH team.
Dataset compiled 2026-08-27 · evidence over inference
Reviewed identity
- Headquarters scope
- EU headquarters
- Headquarters record
- Paris
- Legal entity
- public research body (Institut national de recherche en sciences et technologies du numérique)
- Ownership
- public research
- Founded
- 1967
- 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
- 110M / 335M (base / large)
- Context
- 512 tokens
- Languages
- French
- Licence
- MIT
- Release
- 2019-11
- Type
- text
- Parameters
- 112M
- Context
- 512 tokens
- Languages
- French
- Licence
- MIT
- Release
- 2023
- Type
- text
- Parameters
- base
- Context
- unknown
- Languages
- French
- Licence
- MIT
- Release
- 2024-11
- Type
- text
- Parameters
- base
- Context
- unknown
- Languages
- French
- Licence
- MIT
- Release
- 2024-11
- Type
- text
- Parameters
- 110M
- Context
- 512 tokens
- Languages
- French (biomedical)
- Licence
- MIT
- Release
- 2023-06
Profile note
France's public-research NLP backbone. The CamemBERT line (Inria Paris, ALMAnaCH: Wissam Antoun, Benoît Sagot, Djamé Seddah et al.) is the default French encoder in production across French industry and public administration, and its 2024 v2/av2 refresh remains state of the art for French encoders. Free, MIT, trained on Jean Zay.