EuroEval Swedish NLU - Swerec: leaderboard
The "Swerec" subtask of EuroEval Swedish NLU. Swedish-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: Matthews correlation coefficient (%). Source: euroeval.com. Saturation forecast: Estimated already saturated. 413 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | gemma-3-12B-pt | 81.27 | #686 |
| 2 | gemma-3-27B-pt | 80.97 | #502 |
| 3 | Llama 2 70B Base | 80.95 | #1025 |
| 4 | Llama 3.1 70B | 80.88 | #592 |
| 5 | Gemma 2 9B | 80.82 | #958 |
| 6 | Llama 3.1 8B | 80.44 | #1151 |
| 7 | Qwen 3 235B A22B 2507 Instruct | 80.44 | #300 |
| 8 | Llama 3.1 405B Instruct FP8 | 80.15 | #389 |
| 9 | Qwen 3.5 9B | 80.13 | #379 |
| 10 | Llama 4 Scout Base | 80.1 | #707 |
| 11 | Llama 4 Scout Instruct | 80 | #539 |
| 12 | Yi 1.5 34B | 79.83 | #807 |
| 13 | Gemma 2 27B | 79.58 | #806 |
| 14 | Llama-3.1-Tulu-3-8B-SFT | 79.56 | #985 |
| 15 | Bielik-11B-v2.3-Instruct | 79.52 | #693 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-swedish-nlu-swerec · How It Works · Data refreshed daily, snapshot 2026-10-20.