EuroEval Estonian Summarization - ERR News: leaderboard
EuroEval's Estonian summarization task on the ERR News dataset — models write abstractive summaries of Estonian 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. 150 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.3 70B Instruct | 32.83 | #534 |
| 2 | Gemma 3 27B (IT) | 32.33 | #523 |
| 3 | Mistral Small 3.1 | 32.23 | #610 |
| 4 | Gemma 3 12B (IT) | 32.17 | #664 |
| 5 | Apertus-70B-Instruct-2509 | 31.94 | #820 |
| 6 | Mistral Small 3.2 | 31.88 | #605 |
| 7 | Qwen 3 32B (Non-reasoning) | 31.79 | #443 (Qwen 3 32B) |
| 8 | Qwen 2.5 72B Instruct | 31.66 | #454 |
| 9 | Qwen 3 14B (Non-reasoning) | 31.52 | #537 (Qwen 3 14B) |
| 10 | Gemma 3n E4B (IT) | 31.5 | #749 |
| 11 | GPT-4.1 | 31.32 | #248 |
| 12 | cogito-v1-preview-qwen-14B | 31.32 | |
| 13 | Qwen 3 30B A3B 2507 Instruct | 31.21 | #488 |
| 14 | Gemma 4 31B (IT) | 31 | #176 |
| 15 | Qwen 3 8B (Non-reasoning) | 30.97 | #674 (Qwen 3 8B) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-estonian-summarization-err-news · How It Works · Data refreshed daily, snapshot 2026-10-20.