LLM EU
Menu

Licences in the catalogue

Every model card records a licence identifier and a link to the licence text. This page reads those records back, so you can see which licences the catalogue carries and which models sit under each one.

This page is a catalogue record, not legal advice. No licence text is reproduced or paraphrased here: the link on each row goes to the licence the publisher publishes, and that text governs. Where a licence restricts commercial use, this page says only what the model card already records and points at the card and its limits; where the record does not say what a licence permits, neither does this page.

What the router enforces

Whether a model can be served is not decided by its licence here, and it is not a preference either. Every catalogue model has an endpoint row, and a model the platform cannot serve has one whose kind is catalogue_only. The routing engine rejects that kind before any other rule is considered (packages/router/src/engine.mjs:325), and the model listing refuses it for the same reason (engine.mjs:465). The consequence is structural: a catalogue_only model can never survive candidate filtering, so no policy, no price and no residency choice can route to it. Asking for it by name returns 404 hosted_unavailable with catalogue_only set, not a served model.

Basis: packages/router/src/engine.mjs:325 (the router rejects the kind), engine.mjs:359 (a named catalogue_only model returns hosted_unavailable), engine.mjs:465 (the model listing refuses it too), packages/shared/src/errors.mjs:70 (the 404 body names catalogue_only), scripts/seed.mjs:301 (a model that is not hosted gets exactly that row).

Licences the catalogue records

Computed from the model cards, not from a list kept here. A card with a new licence adds a row the next time this page is rendered.

Licence (SPDX id) Models Licence text Restriction the card records
apache-2.0 open weights
Mistral Small 4 (119B-A6B), Mistral Large 3 (675B), Ministral 3 8B, Devstral Small 2 (24B), Magistral Small 1.2, Voxtral Small 24B, Pixtral 12B, Mistral NeMo 12B Instruct, Mistral 7B Instruct v0.3, EuroLLM-9B-Instruct (2512), EuroLLM-22B-Instruct (2512), EuroLLM-1.7B-Instruct, Qwen3.8-27B, Qwen3-VL-8B-Instruct, Qwen3-8B, Qwen3-Coder-30B-A3B-Instruct, gte-multilingual-reranker-base, gpt-oss-120b, gpt-oss-20b, Whisper large-v3, Gemma 4 31B IT, Gemma 4 26B A4B IT, Gemma 4 E4B IT, Command A+ (05-2026)		 https://www.apache.org/licenses/LICENSE-2.0 · https://ai.google.dev/gemma/docs/gemma_4_license none recorded
mit open weights
DeepSeek-V4-Flash (0731), DeepSeek-V3.2, DeepSeek-R1, DeepSeek-R1-Distill-Qwen-32B, Whisper large-v3-turbo, Phi-4, Phi-4-multimodal-instruct, multilingual-e5-large, GLM-4.7-Flash, GLM-OCR		 https://opensource.org/license/mit none recorded
other open weights
Pharia-1-LLM-7B-control-aligned, Pharia-1-Embedding-4608-control, Llama 4 Scout 17B-16E Instruct, Llama 4 Maverick 17B-128E Instruct, NVIDIA Nemotron 3 Super 120B-A12B, NVIDIA Nemotron 3 Nano 4B, Llama Nemotron Rerank 1B v2, GLM-5.3		 https://huggingface.co/Aleph-Alpha/Pharia-1-LLM-7B-control-aligned-hf · https://www.llama.com/llama4/license/ · https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-nemotron-open-model-license/ · https://openmdw.ai/license/1-1/ · https://huggingface.co/zai-org/GLM-5.3/blob/main/LICENSE
  • Open Aleph License allows non-commercial research and educational use only
  • Llama 4 Community License, not an OSI licence
  • NVIDIA Nemotron Open Model License, a vendor licence rather than an OSI one
  • vendor licence rather than an OSI one
  • bespoke Z.ai licence, which must be read before commercial use
cc-by-nc-4.0 open weights
Teuken 7B Instruct v0.6, Command R (08-2024), Aya Expanse 8B, Aya 23 8B		 https://creativecommons.org/licenses/by-nc/4.0/ · https://cohere.com/c4ai-cc-by-nc-license
  • CC-BY-NC-4.0 forbids commercial use of the released checkpoint
  • CC-BY-NC-4.0 forbids commercial serving of the weights
  • CC-BY-NC-4.0 forbids commercial serving
unknown weights not open
OpenEuroLLM initial model release (planned), OpenEuroLLM final models (planned), LLMEU V2 (announced)		 no licence text URL recorded
  • no parameter count, context window or licence has been published
  • no parameter count, context window or licence has been decided
gemma open weights
EmbeddingGemma 300M		 https://ai.google.dev/gemma/terms
  • Gemma terms of use, not an OSI-approved licence
llama3.1 open weights
Llama 3.1 8B Instruct		 https://www.llama.com/llama3_1/license/
  • Llama 3.1 Community License, not an OSI licence, with a 700M MAU threshold
llama3.3 open weights
Llama 3.3 70B Instruct		 https://www.llama.com/llama3_3/license/
  • Llama 3.3 Community License with a 700M MAU threshold and acceptable-use policy
proprietary weights not open
Codestral 25.08		 https://legal.mistral.ai/ai-governance/models/codestral
  • closed weights: inference only through Mistral or a licensed deployment

Models with no licence permission recorded

These cards record either no licence identifier or closed weights. Nothing here implies permission to serve them, and they are listed so the absence is visible rather than folded into a count.

Not legal advice, and the text governs

LLM EU is not a law firm and this page does not tell you whether a use is permitted. What it does is show the identifier each card records and link the text the publisher publishes. Read that text, and the model's own card, before relying on a licence: the card also carries the limits the catalogue recorded for that model, which is where a restriction on commercial serving is stated first.

Verified — License

Basis: packages/content/models.mjs (license_spdx, license_url, open_weights, limits), packages/router/src/engine.mjs:325, scripts/seed.mjs:292.