Skip to main content
BenchLM
Recommendation

European AI Models in 2026

As of September 25, 2026, the top model in european ai models on the BenchLM leaderboard is Mistral Medium 3.5 128B with a score of 36.1.

Bottom line: Europe's AI model ecosystem is led by Mistral but remains more fragmented than US or Chinese offerings. Mistral Large 3 is the only competitive European model on the leaderboard.

Data verified

Full Rankings (36 models)

1
Mistral Medium 3.5 128B
Mistral·Open Weight·256K

36.1

BenchAlign v5.7

2
Mistral Small 4
Mistral·Open Weight·256K

34.3

BenchAlign v5.7

3
Mistral Large 3
Mistral·Proprietary·128K

34.1

BenchAlign v5.7

4
Mistral Large 2
Mistral·Proprietary·128K

29

BenchAlign v5.7

5
Mistral Medium 3
Mistral·Proprietary·128K

28.3

BenchAlign v5.7

6
Ministral 3 14B
Mistral·Open Weight·128K

25.3

BenchAlign v5.7

7
Mixtral 8x22B Instruct v0.1
Mistral·Open Weight·64K

21.1

BenchAlign v5.7

8
Ministral 3 8B
Mistral·Open Weight·128K

19.3

BenchAlign v5.7

9
Ministral 3 3B
Mistral·Open Weight·128K

18.1

BenchAlign v5.7

10
Mistral 7B v0.3
Mistral·Open Weight·32K

8.1

BenchAlign v5.7

—
Holo2-235B-A22B
H Company·Open Weight·262K

tracked

—
Holo2-30B-A3B
H Company·Open Weight·262K

tracked

—
Holo2-4B
H Company·Open Weight·262K

tracked

—
Holo2-8B
H Company·Open Weight·262K

tracked

—
Holo3-122B-A10B
H Company·Proprietary·64K

tracked

—
Holo3-35B-A3B
H Company·Open Weight·64K

tracked

—
Holo3.1-0.8B
H Company·Open Weight·262K

tracked

—
Holo3.1-35B-A3B
H Company·Open Weight·262K

tracked

—
Holo3.1-35B-A3B-FP8
H Company·Open Weight·262K

tracked

—
Holo3.1-35B-A3B-GGUF
H Company·Open Weight·262K

tracked

—
Holo3.1-35B-A3B-NVFP4
H Company·Open Weight·262K

tracked

—
Holo3.1-4B
H Company·Open Weight·262K

tracked

—
Holo3.1-9B
H Company·Open Weight·262K

tracked

—
Holotron-12B
H Company·Open Weight·128K

tracked

—
Leanstral
Mistral·Open Weight·256K

tracked

—
Leanstral 1.5
Mistral·Open Weight·256K

tracked

—
Ministral 3 14B (Reasoning)
Mistral·Open Weight·128K

tracked

—
Ministral 3 3B (Reasoning)
Mistral·Open Weight·128K

tracked

—
Ministral 3 8B (Reasoning)
Mistral·Open Weight·128K

tracked

—
Mistral 8x7B
Mistral·Open Weight·32K

tracked

—
Mistral 8x7B v0.2
Mistral·Open Weight·32K

tracked

—
Mistral Small 4 (Reasoning)
Mistral·Open Weight·256K

tracked

—
OriOn-Mistral-24B
LightOn·Open Weight·344K

tracked

—
OriOn-Qwen-32B
LightOn·Open Weight·262K

tracked

—
Pharia-1-LLM-7B-control
Aleph Alpha·Open Weight·8K

tracked

—
Pharia-1-LLM-7B-control-aligned
Aleph Alpha·Open Weight·8K

tracked

What changed

Mistral Large 3 only competitive European model — leads on multimodal (67) and math (65).

Mistral Large 2 previous generation, now well behind Large 3.

How to choose

Key Takeaways

The top model is Mistral Medium 3.5 128B by Mistral with a BenchAlign v5.7 score of 36.1 and Estimated evidence.

The best open-weight model is Mistral Medium 3.5 128B at position #1.

36 European models listed, including 10 ranking-eligible and 26 tracked sparse rows.

Score in Context

What these scores mean

European models are ranked by the same overall BenchLM score. This page collects models from European labs for regional comparison.

Known limitations

European model coverage is sparse — most tracked rows outside Mistral have limited benchmark data. H Company, LightOn, and Aleph Alpha are tracked but may lack enough sourced scores to rank.

About this ranking

Last verified: September 25, 2026

European AI models from Mistral, H Company, LightOn, and Aleph Alpha — ranked models first, then tracked sparse rows.

Europe has a credible model ecosystem, but it is more fragmented than China or Korea. This page shows the full tracked European surface on BenchLM: ranking-eligible Mistral rows first, followed by tracked but currently sparse rows from Mistral, H Company, LightOn, and Aleph Alpha.

Unless noted otherwise, ranking surfaces on this page use BenchLM’s provisional leaderboard lane rather than the stricter sourced-only verified leaderboard.

BenchLM currently tracks 36 European models. Only 10 are ranking-eligible today, so this page shows the full tracked set: ranked European rows first, then tracked but benchmark-sparse models below. Right now the ranked subset is led by Mistral Medium 3.5 128B and Mistral Small 4, while the remaining 26 rows are still useful to browse for sovereignty, document AI, and computer-use coverage even when they are not ranking-eligible yet.

The best open-weight option is Mistral Medium 3.5 128B (ranked #1 with a score of 36.1). Open-weight models are highly competitive in this category — self-hosting is a viable alternative to proprietary APIs.

This ranking uses provisional overall weighted scores from the active scoring formula. For detailed model profiles, click any model name above. To compare two specific models head-to-head, use the "vs #" links.

Explore More

Last updated: September 25, 2026

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.