EuroEval Norwegian Summarization - NO Sammendrag: leaderboard
EuroEval's Norwegian summarization task on the NO Sammendrag dataset — models write abstractive summaries of Norwegian news articles, scored with automatic text-overlap metrics, from the EuroEval European language model benchmark suite.
Metric: chrF3++ (%). Source: euroeval.com. Saturation forecast: Around 2035. 155 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.3 70B Instruct | 32.63 | #534 |
| 2 | Mistral Small 3.1 | 31.99 | #610 |
| 3 | Gemma 3 27B (IT) | 31.92 | #523 |
| 4 | Gemma 4 31B (IT) | 31.91 | #176 |
| 5 | Mistral Small 3.2 | 31.9 | #605 |
| 6 | Gemma 3 12B (IT) | 31.82 | #664 |
| 7 | Llama 3.1 8B Instruct | 31.34 | #1035 |
| 8 | Gemma 4 26B A4B (IT) | 31.25 | #297 |
| 9 | Gemini 3 Flash (Preview) (Non-reasoning) | 31.08 | #82 (Gemini 3 Flash (Preview)) |
| 10 | Voxtral-Small-24B-2507 | 31.07 | #581 |
| 11 | EuroLLM-22B-Instruct-2512 | 30.98 | #701 |
| 12 | Claude Sonnet 4.5 (Non-reasoning) | 30.96 | #142 (Claude Sonnet 4.5) |
| 13 | Gemini 3 Pro (Preview) | 30.94 | #66 |
| 14 | cogito-v1-preview-qwen-14B | 30.88 | |
| 15 | Qwen 3 32B (Non-reasoning) | 30.86 | #443 (Qwen 3 32B) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-norwegian-summarization-no-sammendrag · How It Works · Data refreshed daily, snapshot 2026-10-20.