European LLMs in 2026: The Complete Map

Roughly two dozen European organisations have set out to build their own AI models since 2023. Four are dead or dormant, one is being bought by a Canadian company, one is a promise rather than a product, and none cracks the top ten of the main overall boards, where the leaders are all American or Chinese.

This is the full map, and the home of our European large language model (LLM) review series.

Why This Map Follows the Money

Germany’s OpenGPT-X consortium did everything by the book. It delivered Teuken-7B, an LLM that runs in all 24 EU languages, its federal funding ran exactly to schedule, and in early 2025 the project wrapped up on time. The model is still downloadable, but nobody is training the next one.

That is how most European LLM projects end: not in failure, but in a grant that finishes. The cap table tells a blunter story. One company in Europe is trying to fund a frontier lab commercially. Everyone else is running on public grants, academic compute time, or an exit.

So that is how this map is organised. Not by benchmark score, but by who is paying, for how long, and what happens when the money stops.

Every Active Project in One Table

Licences matter here more than parameter counts. “Open” is doing a lot of unearned work in European AI communication, so the column says what the licence actually permits. The projects that have stopped moving are pulled into a separate table below, so this one shows only what is still being built.

The Graveyard: Dead or Dormant

Four projects have effectively stopped. The weights are still downloadable, but nobody is training a successor.

Three Different Bets Under One Word

Read down that table and three separate projects are happening under one word.

The Commercial Bet

Mistral is the only European lab attempting a frontier model on commercial money, and it is doing something unusual with it: it released Mistral Large 3, 41 billion active parameters out of 675 billion total, under Apache 2.0 in December 2025 at $0.50 per million input tokens. That is a genuine open release at frontier scale, and it is the strongest single thing Europe has. The catch, which Mistral’s own communication tends to skip, is that its most capable model is Medium 3.5, and Medium 3.5 is a closed API.

The National Champions

This is the most crowded lane. Poland has two models, one built largely by volunteers and one by the state. Bulgaria’s INSAIT built BgGPT 3.0 on Google’s Gemma 3. Portugal shipped Amália on 1 July for a reported €7M by building on EuroLLM rather than starting from scratch. Hungary’s Racka is a LoRA adaptation of Alibaba’s Qwen-3.

Most national sovereign models are national fine-tunes of somebody else’s base model, and increasingly that base model is Chinese or American. Portugal’s route, adapting an existing European base cheaply, is the honest version of this and probably the smartest one on the table.

The EU Consortia

Here the EU tries to do it collectively. EuroLLM (Unbabel and Edinburgh) released a 22B model in December 2025 and is the one shipping. OpenEuroLLM has published progress and reference models but no flagship. And the newest one, the Domyn-led EUROPA consortium, has been handed EuroHPC compute to build a 400 billion parameter model in all 24 EU languages, which its chief executive says will ship within a year. That is a promise with a compute allocation attached, not a model.

Who Actually Downloads These Models?

We pulled the last-30-day download counts for each project’s headline model from the Hugging Face API on 27 July, and the result rearranges the whole map.

Problems with This Approach

There are four honest caveats with this approach:

• Downloads are not usage. API and self-hosted traffic do not show up, which is why Mistral’s flagship looks small (its 3B Ministral, the one people run locally, did about 700,000 in the same window).
• Smaller models get pulled more than 70B ones, because more people can run them.
• We counted one headline repository per project, so family totals run higher.
• Bulk automated pulls can inflate any single number.

None of that softens the pattern, and one thing sharpens it. Bielik and Llama 4 are gated behind the same access form, so the Polish model’s four-times lead is not an artefact of easier access. It out-pulls Meta’s Llama 4 Maverick eight to one.

Meanwhile the biggest public budgets sit at the bottom. Spain’s ALIA, flagship of a roughly €240M programme, managed 231 downloads in a month. Bulgaria’s BgGPT, 348. Poland’s own state model, PLLuM, is beaten by its volunteer-built compatriot two thousand to one. If public money here is meant to buy adoption, the market has voted, and it did not vote for the ministries.

How Europe Compares to China

On the open-weight benchmarks the field actually watches, Chinese labs run the board. Moonshot released Kimi K3 on 17 July at 2.8 trillion parameters, the largest open-weight model anyone has shipped. Alibaba previewed a 2.4 trillion parameter Qwen two days later, and DeepSeek’s V4 has been live since April. Europe’s largest open release is 675B, and ranks, by Mistral’s own account, second among open non-reasoning models and sixth among open models overall.

Second in a category is a real achievement for a company Mistral’s size, and it should be said plainly. It is also not the same as “Europe has a frontier model”, which is what a lot of European AI communication implies.

The compute gap explains most of it. Mistral raised $830M in debt for a Paris data centre with about 13,800 Nvidia GPUs. Meta guided $115bn to $135bn of capex for 2026 alone. The EU’s answer, the AI Gigafactories programme and the EuroHPC AI Factories network that got Sofia its €90M, is real money that mostly arrives in 2027 and later, the wrong side of the gap it is meant to close.

What to Watch for the Rest of 2026

Three things will decide how this map reads by year end.

• The Cohere and Aleph Alpha deal. Announced 24 April, still unconfirmed three months on. Either the most pragmatic move in European AI or proof the national-champion model does not work.
• Mistral’s reported €20bn round. Bloomberg, Sifted and the Financial Times (Samsung, up to €1bn, 22 July) all describe talks, not a signed round. Anyone quoting €20bn as fact is quoting a negotiation.
• The EUROPA 400B model. Whether it ships by mid-2027 or joins Teuken in the archive. So far only Bielik and Bulgaria’s BgGPT have outlived the grant that started them.

Europe has more working models than the “no AI” line suggests, and fewer than the announcements imply. One commercial frontier attempt, a handful of national models that are actually maintained, two consortia of which one ships, and a graveyard that keeps growing. We test them one by one; the reviews linked here are where the counting gets done.

Author: Akos Szima

This article is for information only and is not financial advice. Figures verified against primary sources on 27 July 2026; download counts pulled from the Hugging Face API the same day; funding rounds described as reported but unclosed are flagged as such.

See Also:

What Is Bielik? Poland’s Sovereign AI Model, Tested (2026)

What Happened to Le Chat? Mistral’s Vibe Rebrand, Tested (2026)