EuroEval Finnish NLU: leaderboard
Finnish-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: NLU Average Score (%). Source: euroeval.com. Saturation forecast: Around March 2027. 382 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 35B A3B FP8 | 73.74 | #362 |
| 2 | Gemma 4 31B (IT) | 71.74 | #176 |
| 3 | Llama 3.1 405B | 70.61 | #466 |
| 4 | Llama 3.1 70B | 70.45 | #592 |
| 5 | Llama 3 70B | 69.49 | #631 |
| 6 | GPT-5 (2025-08-07) (High) | 69.29 | |
| 7 | Gemini 3 Pro (Preview) | 69 | #66 |
| 8 | gemma-3-27B-pt | 69 | #502 |
| 9 | Mistral Small 3.1 | 68.92 | #610 |
| 10 | GPT-5 | 68.64 | #98 |
| 11 | Gemini 3 Flash (Preview) (Thinking) | 68.18 | #82 (Gemini 3 Flash (Preview)) |
| 12 | GPT-5.4 Mini (2026-03-17) (High) | 68.09 | #169 (GPT-5.4 Mini (2026-03-17)) |
| 13 | Gemini 3.1 Flash Lite (Preview) | 67.79 | #194 |
| 14 | Qwen 3 235B A22B 2507 FP8 (Thinking) | 67.67 | #246 (Qwen 3 235B A22B 2507 FP8) |
| 15 | Mistral Small 3.2 | 67.47 | #605 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-finnish-nlu · How It Works · Data refreshed daily, snapshot 2026-10-20.