LLM EU
Menu
Models

Llama Embed Nemotron 8B

NVIDIA

Catalogue only Open weights

Last reviewed:

Sovereignty and hosting

Catalogue only

This model is catalogued for reference. It is not hosted on LLM EU yet, and no first-party endpoint serves it here. Requests to it go to the provider directly, under that provider's own terms.

Summary

Llama Embed Nemotron 8B is published by NVIDIA as open weights with an unpublished parameter count under other. It turns text into vectors for retrieval instead of generating answers. The published configuration records no context window, so this card carries none. It is a base checkpoint without instruction tuning. The card is catalogue-only: LLM EU does not run this model. The publisher states no language coverage on the model page.

Specifications

Provider
NVIDIA (US)
Context window
Not verified: the provider has not documented this claim.
Parameters
Not verified: the provider has not documented this claim.
Modalities
text, embedding
License
other · licence
Open weights
yes
Release date
07 Oct 2025
Tasks
Retrieval (RAG), Multilingual support

Languages

Strong: —

Adequate: —

Strong

    Adequate

      Not verified: the provider has not documented this claim. — a language is listed as strong only where a source claims it; the card links the evidence where one exists.

      When to use it

      Strengths

      • open weights are published on Hugging Face and can be downloaded without a vendor agreement
      • produces embeddings that a European operator can run next to its own vector store

      Limits

      • catalogue only — LLM EU does not run this model, and nothing on this card implies an LLM EU endpoint
      • no evaluation has been run by LLM EU
      • the other licence is a custom agreement whose commercial terms were not established from its published text, so commercial serving is not permitted until a human clears it; the router must not route to it
      • the published configuration records no context window, so the card carries none
      • the publisher states no parameter count in the harvested record
      • the hardware footprint cannot be estimated because no parameter count is published