Evidence-backed European AI register
Artificial Intelligence Laboratory, Institute of Mathematics and Computer Science, University of Latvia
Reviewed identity, roles, model listings, pricing and evidence for Artificial Intelligence Laboratory, Institute of Mathematics and Computer Science, University of Latvia.
Dataset compiled 2026-08-27 · evidence over inference
Reviewed identity
- Headquarters scope
- EU headquarters
- Headquarters record
- Riga
- Legal entity
- laboratory within the Institute of Mathematics and Computer Science, University of Latvia (public research institute, founded 1959)
- Ownership
- public research
- Founded
- 1992
- Confidence
- medium
- 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
- whisper-large-v3-lv-phono
- Type
- asr
- Parameters
- 2B
- Context
- unknown
- Languages
- Latvian
- Licence
- unknown
- Release
- unknown
- lv-deberta-base / lv-roberta-base
- Type
- text
- Parameters
- 0.1B each
- Context
- 512 tokens
- Languages
- Latvian
- Licence
- unknown
- Release
- unknown
- lv-mbert (mini 76.2M / base 0.2B / large 0.4B)
- Type
- text
- Parameters
- 76.2M, 0.2B, 0.4B
- Context
- 512 tokens
- Languages
- Latvian (multilingual BERT adaptations)
- Licence
- unknown
- Release
- unknown
- lv-ner-v1
- Type
- text
- Parameters
- 0.1B
- Context
- 512 tokens
- Languages
- Latvian
- Licence
- unknown
- Release
- unknown
- EuroLLM-9B-Instruct-LVportals-15K
- Type
- text
- Parameters
- 9B — Latvian fine-tune of EuroLLM-9B-Instruct
- Context
- unknown
- Languages
- Latvian
- Licence
- unknown
- Release
- unknown
- Llama3.1-8B-Instruct-LVportals-15K
- Type
- text
- Parameters
- 8B — Latvian fine-tune of Llama 3.1 8B Instruct
- Context
- unknown
- Languages
- Latvian
- Licence
- Llama 3.1 Community License (derivative)
- Release
- unknown
- LVBERT
- Type
- text
- Parameters
- unknown
- Context
- 512 tokens
- Languages
- Latvian
- Licence
- unknown
- Release
- 2020
Profile note
Latvia's academic NLP lab, founded 1992 inside an institute that dates to 1959 — the country's oldest continuous computing research body. Produced LVBERT (2020, the first monolingual Latvian model) and today publishes 23 models and 10 datasets: Latvian encoders, NER, a Latvian-phonological Whisper-large-v3, and Latvian instruction fine-tunes of both EuroLLM-9B and Llama-3.1-8B. It also releases parallel emotion datasets in Latvian, Lithuanian, English and Russian. Complements Tilde: Tilde does the big base model and the commercial layer, AiLab does the task models and evaluation resources. No…