Ranking updated on 09/10/2026
For organizations seeking independence from US or Chinese models, the European open-weights offering is rich in 2026: Mistral AI (🇫🇷), Lucie (🇫🇷 CNRS/LINAGORA), Teuken (🇩🇪 OpenGPT-X), EuroLLM (🇪🇺 EU consortium), CroissantLLM (🇫🇷 bilingual). Ranked by quality + transparency.
24B coding specialist, Apache 2.0. 72.2% SWE-Bench. 256k ctx, FR lab.
ollama run devstral-small2:24b
First open Mistral reasoner. AIME24 70.7%. Based on Small 3.1 + CoT training.
ollama run magistral:24b
June 2025 update for Small 3.1. Half as many infinite generations, improved function calling.
ollama run mistral-small3.2:24b
Excellent quality-to-size ratio at release (early 2025). Competes with 70B models.
ollama run mistral-small:24b
Small 3 enhanced with vision. 128k ctx, Apache 2.0. Replaced by Small 3.2 since June 2025.
ollama run mistral-small3.1:24b
Sovereign French-language LLM, trained on French corpora.
ollama run lucie:7b
8-expert MoE. High quality, but VRAM-hungry.
ollama run mixtral:8x7b
The classic French model. Fast, versatile, and an excellent base to get started.
ollama run mistral:7b-instruct
| Rank | Model | Params | Q4 VRAM | Context | License |
|---|---|---|---|---|---|
| #1 | Devstral Small 2 24B | 24B | 14 GB | 256 000 | Apache 2.0 |
| #2 | Magistral Small 24B | 24B | 14 GB | 128 000 | Apache 2.0 |
| #3 | Mistral Small 3.2 24B | 24B | 14 GB | 128 000 | Apache 2.0 |
| #4 | Mistral Small 3 | 24B | 14 GB | 32 768 | Apache 2.0 |
| #5 | Mistral Small 3.1 24B | 24B | 14 GB | 128 000 | Apache 2.0 |
| #6 | Lucie 7B | 7B | 5 GB | 4 096 | Apache 2.0 |
| #7 | Mixtral 8x7B | 47B | 26 GB | 32 768 | Apache 2.0 |
| #8 | Mistral 7B Instruct | 7B | 5 GB | 32 768 | Apache 2.0 |
Your private, free ChatGPT on your machine in 1 hour — LM Studio, Ollama, Open WebUI, your documents, no cloud.
Filter: French origin OR a known European author (Mistral, OpenLLM-France, OpenGPT-X, EuroLLM, etc.). Sovereignty bonus for 100% French models with a transparent corpus (Lucie).
Criteria considered:
The scoring is fully transparent: see our methodology for details on VRAM/tokens/sec calculations.
Why a sovereign European LLM?
Data sovereignty (no transfers outside the EU), easier GDPR compliance, alignment with European values, and resilience to changes in US/China trade policy. For defense, healthcare, and government agencies, this is often a mandatory criterion.
Is Mistral AI really European?
Yes — a French company based in Paris, founded by former Meta and DeepMind employees. The “Open” models (7B, Small, Mixtral) are licensed under Apache 2.0. Only the “Commercial” models (Large) are proprietary.
Is Lucie 7B at the level of Mistral 7B?
No, Lucie 7B is less versatile in English and limited to 4k context tokens. But it is the only 100% sovereign model with a fully documented and public training corpus (Jean Zay / CNRS). View the comparison.
Is there a European multilingual LLM that covers all 24 EU languages?
Yes — EuroLLM 9B et Teuken 7B are explicitly trained on the 24 official EU languages. They are weaker in English than Mistral but excellent for cross-European use cases.
Learn more with our detailed head-to-head matchups of the finalists: