Mistral 7B
Mistral AI · France
- Params
- 7B
- Context
- 32,768
- Min VRAM
- 8 GB
- GGUF
- Yes
Fast inference, strong reasoning, multilingual support, open weights.
Region slug: europeEurope (FR / DE / Spain / Nordic / Italy / Poland / NL / EU-consortium)
AI models, vendors, and hardware buying realities from Europe (FR / DE / Spain / Nordic / Italy / Poland / NL / EU-consortium). Curated by myaihardware.com.
32
Models
9
Vendors
,
Languages
Why Europe (FR / DE / Spain / Nordic / Italy / Poland / NL / EU-consortium) matters for AI builders
For AI builders in Europe, hardware access is a bottleneck — NVIDIA H100s are scarce and costly, while AMD MI250/MI300X systems are easier to provision through local CSPs. The dominant model families driving inference and fine-tuning here are Mistral, Llama, and the open-weight Qwen variants, not US-centric proprietary labs. Underrated: Europe's strong data-protection laws create a real demand for on-premise inference stacks, making local builders early adopters of hybrid or edge-hardware solutions that the rest of the market will chase later.
Models (32 of 32)
Filter by vendor, license, or parameter range.
Open
Mixtral 8x7B
Mistral AI · France
- Params
- 8x7B
- Context
- 32,768
- Min VRAM
- 24 GB
- GGUF
- Yes
High performance at fraction of full-parameter cost, excellent multilingual reasoning.
Region slug: europeMixtral 8x22B
Mistral AI · France
- Params
- 8x22B
- Context
- 65,536
- Min VRAM
- 48 GB
- GGUF
- Yes
Strong reasoning, long context, high throughput.
Region slug: europeMistral Large 2
Mistral AI · France
- Params
- 123B
- Context
- 128,000
- Min VRAM
- 80 GB
- GGUF
- Yes
Top-tier reasoning, multilingual performance, long context, instruction following.
Region slug: europeMistral Small
Mistral AI · France
- Params
- 22B
- Context
- 32,768
- Min VRAM
- 16 GB
- GGUF
- Yes
Balanced efficiency, strong multilingual chat, affordable inference.
Region slug: europeCodestral 22B
Mistral AI · France
- Params
- 22B
- Context
- 32,768
- Min VRAM
- 16 GB
- GGUF
- Yes
Excellent code generation, fill-in-the-middle, multi-language fluency.
Region slug: europeMistral NeMo 12B
Mistral AI / NVIDIA · France
- Params
- 12B
- Context
- 128,000
- Min VRAM
- 12 GB
- GGUF
- Yes
Long context, strong multilingual, instruction following, compact footprint.
Region slug: europePixtral 12B
Mistral AI · France
- Params
- 12B
- Context
- 128,000
- Min VRAM
- 16 GB
- GGUF
- Yes
Multimodal understanding, image captioning, visual QA, strong language backbone.
Region slug: europeMinistral 3B
Mistral AI · France
- Params
- 3B
- Context
- 128,000
- Min VRAM
- 4 GB
- GGUF
- Yes
Extremely lightweight, fast inference, long context, low power.
Region slug: europeMinistral 8B
Mistral AI · France
- Params
- 8B
- Context
- 128,000
- Min VRAM
- 8 GB
- GGUF
- Yes
Good balance of size and capability, long context, resource-efficient.
Region slug: europeMathstral 7B
Mistral AI · France
- Params
- 7B
- Context
- 32,768
- Min VRAM
- 8 GB
- GGUF
- Yes
Specialised for mathematics, strong symbolic reasoning, theorem proving.
Region slug: europeMistral Embed
Mistral AI · France
- Params
- 7B
- Context
- 8,192
- Min VRAM
- 8 GB
- GGUF
- No
High-quality multilingual text embeddings, retrieval performance.
Region slug: europeLucie-7B
LinAGora · France
- Params
- 7B
- Context
- 4,096
- Min VRAM
- 14 GB
- GGUF
- Yes
Strong French fluency, transparent training data, open license, suitable for French public sector and education use cases.
Region slug: europeCroissantLLM-1.3B
CentraleSupélec · France
- Params
- 1.3B
- Context
- 2,048
- Min VRAM
- 3 GB
- GGUF
- Yes
Efficient tiny bilingual model, fully open research artifact, transparent training recipe, runs on CPU.
Region slug: europeEuroLLM-9B
EU-consortium (Unbabel, TU Berlin, UT Austin, etc.) · EU-consortium
- Params
- 9B
- Context
- 8,192
- Min VRAM
- 18 GB
- GGUF
- Yes
Multilingual across all EU languages, open weights, strong translation and cross-lingual transfer.
Region slug: europeEuroLLM-22B-Preview
EU-consortium (Unbabel, TU Berlin, UT Austin, etc.) · EU-consortium
- Params
- 22B
- Context
- 16,384
- Min VRAM
- 44 GB
- GGUF
- Yes
Larger multilingual capacity, extended context, improved reasoning across EU languages.
Region slug: europePharia-1-LLM-7B-base
Aleph Alpha · Germany
- Params
- 7B
- Context
- 8,192
- Min VRAM
- 16 GB
- GGUF
- No
German-engineered foundation model, strong European language support, compliance-first design.
Region slug: europePharia-1-LLM-7B-instruct
Aleph Alpha · Germany
- Params
- 7B
- Context
- 8,192
- Min VRAM
- 16 GB
- GGUF
- No
Instruction-tuned for enterprise tasks, strong German-language performance, compliant deployment.
Region slug: europeTeuken-7B-instruct
OpenGPT-X / Fraunhofer IAIS · Germany
- Params
- 7B
- Context
- 4,096
- Min VRAM
- 14 GB
- GGUF
- Yes
Fully open European multilingual LLM, transparent training, strong on 20+ EU languages.
Region slug: europeALIA-40B
Barcelona Supercomputing Center (BSC) · Spain
- Params
- 40B
- Context
- 4,096
- Min VRAM
- 80 GB
- GGUF
- Yes
Largest open Iberian language model, strong on Catalan, Basque, Galician; transparent training.
Region slug: europeSalamandra-7B
Barcelona Supercomputing Center (BSC) · Spain
- Params
- 7B
- Context
- 8,192
- Min VRAM
- 14 GB
- GGUF
- Yes
Compact multilingual model for Romance languages, strong Spanish and Catalan fluency, open weights.
Region slug: europeSalamandra-40B
Barcelona Supercomputing Center (BSC) · Spain
- Params
- 40B
- Context
- 8,192
- Min VRAM
- 80 GB
- GGUF
- Yes
Large-scale Romance language model, deep reasoning, strong across all Iberian languages.
Region slug: europeViking-7B
Silo AI (now AMD) · Finland
- Params
- 7B
- Context
- 4,096
- Min VRAM
- 14 GB
- GGUF
- Yes
Strong Nordic language performance, compact size, open weights, runs on consumer hardware.
Region slug: europeViking-13B
Silo AI (now AMD) · Finland
- Params
- 13B
- Context
- 4,096
- Min VRAM
- 26 GB
- GGUF
- Yes
Balanced Nordic multilingual model, better reasoning than 7B, still consumer-GPU friendly.
Region slug: europeViking-33B
Silo AI (now AMD) · Finland
- Params
- 33B
- Context
- 8,192
- Min VRAM
- 66 GB
- GGUF
- Yes
Most capable Nordic LLM, strong multilingual reasoning, extended context for document tasks.
Region slug: europePoro-34B
Silo AI (now AMD) · Finland
- Params
- 34B
- Context
- 8,192
- Min VRAM
- 68 GB
- GGUF
- Yes
Flagship Nordic-Baltic model, strong Finnish fluency, broad language coverage in region.
Region slug: europeGEITje-7B-ultra
Rijgersberg / Dutch community · Netherlands
- Params
- 7B
- Context
- 8,192
- Min VRAM
- 14 GB
- GGUF
- Yes
Exceptional Dutch fluency, culturally aware, fine-tuned for Dutch tasks, strong instruction following.
Region slug: europeMinerva-7B-instruct
Sapienza University of Rome · Italy
- Params
- 7B
- Context
- 4,096
- Min VRAM
- 14 GB
- GGUF
- Yes
Leading open Italian LLM, strong academic foundation, transparent and reproducible training.
Region slug: europeBielik-7B-Instruct-v0.1
SpeakLeash · Poland
- Params
- 7B
- Context
- 4,096
- Min VRAM
- 14 GB
- GGUF
- Yes
Strong Polish language performance, culturally tuned, good instruction following.
Region slug: europeBielik-11B-v2
SpeakLeash · Poland
- Params
- 11B
- Context
- 8,192
- Min VRAM
- 22 GB
- GGUF
- Yes
Improved multilingual coverage, extended context, stronger reasoning than 7B predecessor.
Region slug: europePLLuM-12B
PLLuM Consortium (Poznań University of Technology et al.) · Poland
- Params
- 12B
- Context
- 8,192
- Min VRAM
- 24 GB
- GGUF
- Yes
Fully open Polish LLM with permissive license, strong fluency, government-backed initiative.
Region slug: europePULI-GPTrio
Hungarian Research Centre for Linguistics / SZTAKI · Hungary
- Params
- 7B
- Context
- 4,096
- Min VRAM
- 14 GB
- GGUF
- Yes
Dedicated Hungarian LLM, strong on agglutinative morphology, open and permissive license.
Region slug: europeNotable vendors
Mistral AIAleph AlphaOpenGPT-X / Fraunhofer IAISBarcelona Supercomputing Center (BSC)Silo AI (AMD)LinAGoraSpeakLeashSapienza University of RomeEuroLLM consortium
Run benchmarks
llama.cpp, MLPerf, and inference leaderboards.
VRAM calculator
Will this model fit on your GPU?
Hardware guides
Local LLM setup, homelab, mini-PC builds.
Methodology
How we benchmark, verify ASINs, and source data.
Beginner's guide
New to local AI? Start here.
Sourcing & commerce disclosure
Model metadata on this page is compiled from each vendor's public HuggingFace repo, model card, or research paper at time of indexing. Performance figures (KMMLU, MMLU, HAERAE, AraBench, IndicGEM, etc.) link directly to source benchmarks where available; missing or partial values are shown as ", ".
Outbound hardware links (GPUs, mini-PCs, accelerators) are Amazon Associate links using store ID fredoline-20. As an Amazon Associate we earn from qualifying purchases. Prices and availability shown elsewhere on the site are snapshots, not live quotes — verify on Amazon before purchasing. See About / disclosures for the full policy.