Evidence-backed European AI register
deepset
Reviewed identity, roles, model listings, pricing and evidence for deepset.
Dataset compiled 2026-08-27 · evidence over inference
Reviewed identity
- Headquarters scope
- EU headquarters
- Headquarters record
- Berlin
- Legal entity
- GmbH
- Ownership
- VC-backed private
- Founded
- 2018
- 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
- embedding
- Parameters
- 335M
- Context
- 512 tokens
- Languages
- German + English
- Licence
- Apache-2.0
- Release
- 2024-09
- deepset German BERT / German ELECTRA / gelectra, roberta-base-squad2 and the deepset QA model family
- Type
- text
- Parameters
- 110M–355M class
- Context
- 512 tokens
- Languages
- German, English
- Licence
- MIT / Apache-2.0 / CC-BY-4.0 depending on repo
- Release
- 2019–2023
Profile note
Maintainer of Haystack, the most widely used European open-source RAG/agent orchestration framework, and publisher of the German-language BERT/QA models that were the de facto standard before LLMs. Now positions as a sovereign agent platform. It is a framework and platform vendor first; the models it trains are embeddings and encoders, not generative LLMs. Pricing above the free Studio tier is fully gated.