Introduction
I have written about Europe’s AI sovereignty problem four times now (Green 2026a, 2026b), and every time I land in the same uncomfortable place. The two pillars Europe leans on (the closed US frontier and the open Chinese weights frontier) are both rotting, and the continent I live in still has no pillar of its own.
The snapshot date above the table is the version. It changes when a frontier model lands, the AA index changes version, or a set of weights moves behind an endpoint.
The two pillars
Europe runs its frontier AI on two pillars we do not control. Pillar one is the closed US frontier (Anthropic, OpenAI, Meta, xAI, Google) accessed through APIs whose terms can change overnight based on a government we have nothing to do with. Pillar two is the open Chinese weights frontier (Kimi, Xiaomi’s MiMo, DeepSeek, the Qwen open line, and the GLM line) which ships real weights today but ships them from a jurisdiction consulting on restricting overseas access to its most advanced models (The Next Web 2026). Both pillars can be pulled. We saw pillar two wobble in slow motion when Qwen3.8-Max launched API-only with weights “promised within days”, watched it partly reset on 2026-08-13 when the 2.4T-A95B weights landed on HuggingFace, watched it wobble again when Z.ai shipped GLM-5.3 API-first and delayed the weights over the model’s own cyber capability, and then watched the promise kept in full: the Flash tier landed open on 2026-08-26, and the full 5.3 weights landed on 2026-08-28, two weeks to the day after the API launch, on the schedule Z.ai named up front (Z AI 2026a, 2026b). We saw pillar one’s political risk go from theoretical to concrete when Anthropic tried corporate self-regulation for military use and got punished so publicly that no rational actor will try it again (Anthropic 2026a). Europe has its own version of the rented pillar. Spain’s Multiverse Computing serves a compressed, proprietary copy of GLM-5.2 from its own API.
What changed
Two launches moved the board this snapshot, and one of them is European.
Mistral launched a public preview of Mistral Large 4 on 2026-10-06 (Mistral AI 2026). It grades 38.4, the best European score this board has recorded, 11.7 points above Multiverse’s Quasar and 24.2 above Mistral Medium 3.5 (Artificial Analysis 2026g). Europe’s gap to the top shrank from 30.9 points to 19.2. It is Mistral’s own model, a 1-trillion-parameter natively multimodal mixture-of-experts with 49 billion active, trained from scratch on 3,800 Nvidia Grace Blackwell GPUs in Mistral’s own European datacenters. Artificial Analysis calls it the most intelligent model from outside the US and China. It still trails four Chinese open-weights rows on the board, and its own row is still a date. The preview is API-only, with the weights promised for the end of October. That is the StepFun pattern, from a European lab.
Google came back a week earlier, with a model Europe can’t use. Gemini 4 Argon, announced 2026-09-30, grades 52.6, 0.1 behind GPT-6 Astra (Google 2026). Google is rolling it out only to partners in its Fairwind Program, who may give access only to internal cybersecurity, incident response and penetration testing teams (Google DeepMind 2026). The frontier window now holds four labs, all American and closed, and the third-best model on the board is the first row with a “Restricted” verdict.
The rest of the top held. Claude Opus 5.5 still leads at 57.6, 4.9 points ahead of Astra (Anthropic 2026b; Artificial Analysis 2026f). Anthropic and OpenAI shipped cheaper tiers under their flagships since the last snapshot (Claude Sonnet 5.5 at 56.0, GPT-6.1 Sol at 51.8).
Below the window, no score moved. MiMo-V2.6-Pro is still the best open model at 46.3, 11.3 points under the top (Xiaomi MiMo 2026). The Step 5 weights are due 2026-10-15 (StepFun 2026). The Qwen Max hybrid is nine weeks past “within days”, and the closed 0902 endpoint still grades 5.5 points above the open 2.4T (Alibaba 2026b). DeepSeek V4.1 Pro has not shipped (DeepSeek 2026a). One new Chinese lab graded near the window: InclusionAI, with Ling 3.1 Flash at 41.1 on 2026-10-01. The index is still on v4.3.2, so every score on this page compares directly with the last snapshot.
The scoreboard
Snapshot date: 2026-10-07. Scores from the Artificial Analysis Intelligence Index v4.3, pulled from their Data API (free tier) (Artificial Analysis 2026d, 2026c). Each row is the top reasoning variant of that lab’s frontier model, selected automatically: a lab enters when its best model is within ten points of the board’s top score, European labs enter regardless of score, and the curated open-line rows carry the open-weights argument. The Δ top column is each row’s distance from the board’s best score. Treat every number as a point-in-time reading. The index was rebuilt twice between 2026-08-28 and 2026-09-15, so scores on this page compare with the 2026-09-15 snapshot (v4.3, patched to v4.3.2 on 2026-09-19) and with nothing before it. The Δ top column and the ordering are the quantities that compare across all of them.
| Rank | Lab | Region | Latest frontier model | Open? | AA index | Δ top | EU access | Verdict |
|---|---|---|---|---|---|---|---|---|
| 1 | Anthropic | US | Claude Opus 5.5 | Proprietary | 57.6 | 0.0 | API | Revocable (ToS and political risk) |
| 2 | OpenAI | US | GPT-6 Astra | Proprietary | 52.7 | −4.9 | API | Revocable |
| 3 | US | Gemini 4 Argon | Proprietary | 52.6 | −5.0 | None (Fairwind partners only) | Restricted (Fairwind Program cyber defenders only) | |
| 4 | Meta | US | Muse Spark 1.3 | Proprietary | 48.1 | −9.5 | API | Revocable (Meta closed its frontier) |
| 5 | Xiaomi | China | MiMo-V2.6-Pro | Open weights (MIT, shipped 2026-09-21) | 46.3 | −11.3 | Self-host | Revocable (Beijing consulting export limits) |
| 6 | Z.ai | China | GLM-5.3 | Open weights (custom permissive license, shipped 2026-08-28) | 44.8 | −12.8 | Self-host | Revocable (Beijing consulting export limits) |
| 7 | StepFun | China | Step 5 Preview | API-only preview; weights promised for 2026-10-15 | 43.7 | −13.9 | API | Revocable (weights promised, not shipped) |
| 8 | Moonshot | China | Kimi K3 | Open weights | 43.6 | −14.0 | Self-host | Revocable (Beijing consulting export limits) |
| 9 | Alibaba (Flash-Next open) | China | Qwen3.8-Flash-Next | Open weights (Qwen community license) | 39.8 | −17.8 | Self-host | Revocable (Beijing consulting export limits) |
| 10 | Mistral | EU | Mistral Large 4 Preview | API-only preview; weights promised end of October 2026 | 38.4 | −19.2 | API | EU pillar, trained in Europe, weights promised end of October |
| 11 | Alibaba (27B open) | China | Qwen3.8-27B | Open weights (Apache 2.0) | 33.7 | −23.9 | Self-host | Revocable (Beijing consulting export limits) |
| 12 | Multiverse Computing | EU | Quasar 438B | Proprietary (AA: based on GLM-5.2) | 26.7 | −30.9 | API | EU pillar, under construction, far from frontier (AA: GLM-5.2 derivative) |
| 13 | OpenAI (open line) | US | gpt-oss-120b | Open weights | 11.6 | −46.0 | Self-host | Stable, but far from frontier |
| 14 | Swiss AI Initiative | Europe (non-EU) | Apertus 70B Instruct | Open weights | 5.1 | −52.5 | Self-host | European, outside EU jurisdiction |
| n/a | EUROPA / Domyn | EU | In training | TBD | TBD | TBD | EU native | Under construction, >400B params, 24 EU languages |
The Qwen family still spans four openness states: a closed Max endpoint (45.4 for the 0902 refresh, below the window and not a row), an open 2.4T sibling (39.9), an open Flash-Next architecture preview on the Qwen community license (39.8), and an Apache 2.0 consumer 27B (33.7), with the Max hybrid nine weeks past its launch-day promise (Alibaba 2026b, 2026a). The Multiverse cell is proprietary, based on GLM-5.2, and served from Spain (Artificial Analysis 2026h). OpenAI’s gpt-oss-120b, the only American open-weights row, grades 11.6, which puts it in research-artifact territory.
How to read it
The AA index moves week to week as it re-aggregates. It was rebuilt twice in early September and has run on the v4.3.2 patch since 2026-09-19 (Artificial Analysis 2026e). Intelligence Index v4.3 aggregates ten evaluations (AA-Briefcase, GDPval-AA v2, AutomationBench-AA, Terminal-Bench v4.0, SciCode, Humanity’s Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1), and roughly 40 percent of the composite is now graded on private test sets (Artificial Analysis 2026a, 2026b). A model can gain or lose a point or two without a new release, just because the index re-sampled. It can lose a lot more when the yardstick gets replaced. So a point or two between the rows at the top of the board is noise. The signal is which labs are within shouting distance of the frontier top, and which access regime they ship under. The frontier window now closes after row 4, Muse Spark 1.3 at 48.1. Grok 4.7 (46.4), MiMo-V2.6-Pro (46.3), Qwen3.8-Max-0902 (45.4), Ling 3.1 Flash (41.1) and DeepSeek (39.5) all sit below it.
“Open?” is the column that matters most for Europe, and it is the one most likely to lie to you. The MiMo cell is a receipt: MIT weights on HuggingFace the day the model was announced (Xiaomi MiMo 2026). The Step 5 cell is a promise with a date, 2026-10-15, and no license yet (StepFun 2026). The Mistral cell is the same kind of promise from a European lab, weights by the end of October (Mistral AI 2026). The GLM-5.3 cell shows what a kept promise looks like: the Flash sibling open under MIT since 2026-08-26, and the full 5.3 weights on HuggingFace on 2026-08-28, on the two-week schedule Z.ai announced up front (Z AI 2026a, 2026b). The Qwen cell is still a promise: the Max hybrid was promised “within days” on 2026-08-03, and nine weeks later the model card is still not on HuggingFace, while everything below the Max line (2.4T, Flash-Next, 27B) is open and a dated API refresh shipped instead (Alibaba 2026b). Alibaba fenced its Max tier across two generations (Qwen3.7-Max proprietary, Qwen3.8-Max paid preview), and the 2026-08-13 open-weighting of the 2.4T-A95B broke the pattern for everything except the top. The 27B cell has no promise and no asterisk: Apache 2.0, on HuggingFace, downloadable today.
“EU access” comes in three kinds now. API access means you can call the model today and you cannot run it yourself tomorrow if the provider decides you should not. Self-host means you have the weights on hardware you control, and the only thing that can take that away is a license change you can see coming or a jurisdiction deciding to restrict exports. The third kind arrived with Gemini 4 Argon. Unless you are one of Google’s Fairwind partners, you get no access, and partners get it only for their internal security teams (Google DeepMind 2026).
Where Europe sits
Mistral is Europe’s best-graded lab again, with Large 4 Preview ranked #70 among all graded variants (Artificial Analysis 2026g). The ASML-led 1.7 billion euro Series C put real compute behind the company (CNBC 2025), and Large 4 is the first Mistral model on this board within twenty points of the top. The frontier-tier question (does Mistral ship a model that sits in the top three of the AA index, ever) is still open. Until the weights land at the end of October, Europe’s best model is one Europe can call and not yet run.
Multiverse Computing drops to second. Quasar 438B (26.7, ranked #138) is proprietary, served from Multiverse’s own CompactifAI API, and, by Multiverse’s own technical note, “a compressed model built from GLM-5.2”, the Chinese open-weights model, pruned with their quantum-inspired method (Multiverse Computing 2026; Artificial Analysis 2026h). At launch Multiverse billed it as Europe’s leading AI model. Five weeks later Mistral’s own model passed it. The open GLM-5.2 original still leads the compressed copy 33.7 to 26.7, and the original is free to self-host while the copy is API-only. Their previous graded model, HyperNova 60B, is the same pattern: AA labels it “based on gpt-oss-120b”, the American open-weights artifact. Both of Multiverse’s graded rows are compressions of other labs’ open models. That is a real business, and it is a long way from a frontier.
The EUROPA consortium led by Domyn is the other one to watch, and it is unchanged. The EU Frontier AI Grand Challenge is funding a >400 billion parameter model trained on all 24 EU languages on a 6,000-chip Nvidia Blackwell cluster (European Commission 2026). It is still the only announced European project that names the frontier tier as its goal. It is also not a frontier model yet, and the gap between “we are training a 400B model” and “we have a model in the top three of the AA index” is exactly the gap Europe has been failing to close for two years.
Apertus 70B (Swiss AI Initiative, 5.1, ranked #652) is real, downloadable, European, and 52.5 points off the frontier, 33.3 points behind Europe’s best.
There is still no European row in the top half of the table. Mistral Large 4 is row 10 of 14.
What would change the table
These would move the table beyond normal score drift:
- Mistral ships the Large 4 weights by the end of October. Europe gets its first self-hostable model within twenty points of the top, or a European promise joins Qwen Max on the list of promises gone quiet (Mistral AI 2026).
- StepFun ships the Step 5 weights on 2026-10-15. The row flips from a date to a receipt, or it joins Qwen Max as a promise gone quiet (StepFun 2026).
- DeepSeek V4.1 Pro ships. DeepSeek’s best graded model is still V4.1-Flash, open weights under MIT, shipped 2026-09-10, at 39.5 (DeepSeek 2026b). V4.1 Pro has not shipped, and V4 Pro stays on the API for now (DeepSeek 2026a). If V4.1 Pro lands anywhere near the window, DeepSeek is back.
- Qwen3.8-Max hybrid weights actually ship. Nine weeks and one dated API refresh later, still no model card, and the refresh now grades 5.5 points above the open 2.4T. Z.ai has now kept a dated weights promise; Alibaba’s is the standing counterexample. I will believe the hybrid when the model card lands on HuggingFace.
- Qwen4 ships. Alibaba previewed it as in training at Apsara on 2026-09-22, with no dates (Alizila 2026). Flash-Next is still the only open preview of the Qwen4 architecture. If Qwen4 lands as a Max-class open drop, the top of the board is in play. If it lands API-only with open siblings below, the Qwen pattern repeats at the next generation.
- Beijing formalizes the overseas-access restrictions it has been consulting on. The reporting now says the Ministry of Commerce is drafting a tiered export regime that would treat Qwen and DeepSeek weights the way Washington treats advanced chips (The Next Web 2026). Several Chinese rows flip their verdict from “Revocable” to “Restricted” and pillar two gets measurably weaker for Europe.
- A US provider changes terms in a way that breaks EU access (region lock, use-case restriction, government-mandated cutoff). The corresponding US row gets a date stamped on its verdict.
- Google opens Gemini 4 Argon beyond Fairwind. The row’s EU access changes from none to API, and the board’s only restricted row becomes an ordinary revocable one.
- EUROPA publishes a model, or Mistral lands one in the AA index top ten. A row leaves or joins the board.
- Meta re-opens a frontier-tier Muse Spark with open weights, or OpenAI ships an open-weights model above 40. The US regains an open-weights frontier leg, and the two-pillar story gets a third leg I did not expect.
- A non-US lab re-enters the frontier window. Right now the window holds four American closed labs and nothing else. MiMo-V2.6-Pro sits 1.3 points under the line.
Conclusion
Every time a frontier model lands, the good news is about someone else’s model and the bad news is about someone else’s decision. Europe’s role is to read the announcement and adjust its compliance documentation.
Eleven of the fourteen scored rows are non-European, and every row above Europe’s best is a model Europe uses on terms set outside Europe, one of them a model Europe can’t use at all. The good news this round is European: Mistral trained a 1-trillion-parameter model in its own datacenters, and it is the closest a European lab has come to the top of this board. The bad news is that the open frontier above it is Chinese, and Mistral’s weights are a promise until the end of October.
Z.ai named a date for the GLM-5.3 weights and kept it. Xiaomi shipped MiMo’s weights with no promise at all. StepFun named 2026-10-15, and Mistral named the end of October. Alibaba put a new date on a closed API and let the hybrid promise run past nine weeks, while the closed endpoint pulled 5.5 points ahead of the open one. Each of those weights ships when the lab that trained it decides, and Mistral’s is the only date Europe has any say in. The closed US frontier added Google this round, behind a cyber-defender access program, and the rest of it is revocable, as we watched the political risk go concrete in March. The European pillar is under construction, and this round its best model is Mistral’s own, trained in Europe. Of the three pillars, it is the only one Europe controls.