EuroEval Ukrainian Summarization - LR SUM UK: leaderboard
EuroEval's Ukrainian summarization task on the LR SUM UK dataset — models write abstractive summaries of Ukrainian news articles, scored with automatic text-overlap metrics, from the EuroEval European language model benchmark suite.
Metric: chrF3++ (%). Source: euroeval.com. Saturation forecast: Not forecast. 146 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Mistral Small 3.1 | 32.26 | #610 |
| 2 | Qwen 2.5 72B Instruct | 31.7 | #454 |
| 3 | Granite 4.0 Micro | 31.58 | #882 |
| 4 | Qwen 3 4B (Non-reasoning) | 31.56 | #829 (Qwen 3 4B) |
| 5 | occiglot-7B-eu5-instruct | 31.47 | |
| 6 | Apertus-70B-Instruct-2509 | 31.41 | #820 |
| 7 | Mistral Small 3.2 | 31.38 | #605 |
| 8 | Qwen 3 8B (Non-reasoning) | 31.21 | #674 (Qwen 3 8B) |
| 9 | Voxtral-Small-24B-2507 | 31.2 | #581 |
| 10 | Qwen 3 14B (Non-reasoning) | 31.19 | #537 (Qwen 3 14B) |
| 11 | Qwen 3 32B (Non-reasoning) | 31.12 | #443 (Qwen 3 32B) |
| 12 | EuroLLM-9B-Instruct | 31.12 | #1054 |
| 13 | granite-4.0-h-micro | 31.03 | #1088 |
| 14 | Qwen 3 1.7B (Non-reasoning) | 30.85 | #1196 (Qwen 3 1.7B) |
| 15 | Qwen 3 30B A3B 2507 Instruct | 30.72 | #488 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-ukrainian-summarization-lr-sum-uk · How It Works · Data refreshed daily, snapshot 2026-10-20.