EuroEval Icelandic Summarization - RRN: leaderboard
EuroEval's Icelandic summarization task on the RRN dataset — models write abstractive summaries of Icelandic news articles, scored with automatic text-overlap metrics, from the EuroEval European language model benchmark suite.
Metric: chrF3++ (%). Source: euroeval.com. Saturation forecast: Around 2034. 155 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Apertus-70B-Instruct-2509 | 40.02 | #820 |
| 2 | Mistral Small 3.1 | 39.67 | #610 |
| 3 | Mistral Small 3.2 | 39.58 | #605 |
| 4 | Voxtral-Small-24B-2507 | 38.88 | #581 |
| 5 | Qwen 3 14B (Non-reasoning) | 38.85 | #537 (Qwen 3 14B) |
| 6 | Gemma 3n E4B (IT) | 38.47 | #749 |
| 7 | GPT-4.1 | 38.42 | #248 |
| 8 | Gemma 3 27B (IT) | 38.35 | #523 |
| 9 | Gemma 3 12B (IT) | 38.14 | #664 |
| 10 | Qwen 3 8B (Non-reasoning) | 37.96 | #674 (Qwen 3 8B) |
| 11 | Llama 3.3 70B Instruct | 37.93 | #534 |
| 12 | Qwen 2.5 72B Instruct | 37.9 | #454 |
| 13 | Llama 3.1 8B Instruct | 37.88 | #1035 |
| 14 | Qwen 3 4B (Non-reasoning) | 37.83 | #829 (Qwen 3 4B) |
| 15 | Gemini 3 Pro (Preview) | 37.72 | #66 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-icelandic-summarization-rrn · How It Works · Data refreshed daily, snapshot 2026-10-20.