EuroEval Italian Summarization - Ilpost SUM: leaderboard
EuroEval's Italian summarization task on the Ilpost SUM dataset — models write abstractive summaries of Italian 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. 151 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Mistral Small 3.1 | 39.2 | #610 |
| 2 | Gemma 4 E2B (IT) | 38.97 | #781 |
| 3 | gemma-4-E4B-it | 38.85 | #566 |
| 4 | GPT-4.1 | 38.72 | #248 |
| 5 | Apertus-70B-Instruct-2509 | 38.7 | #820 |
| 6 | Bielik-11B-v2.3-Instruct | 38.66 | #693 |
| 7 | Gemma 3 12B (IT) | 38.63 | #664 |
| 8 | Qwen 3 14B (Non-reasoning) | 38.58 | #537 (Qwen 3 14B) |
| 9 | Qwen 3 8B (Non-reasoning) | 38.55 | #674 (Qwen 3 8B) |
| 10 | Qwen 3.5 35B A3B Base | 38.54 | #310 |
| 11 | Gemma 3 27B (IT) | 38.42 | #523 |
| 12 | Gemma 4 26B A4B (IT) | 38.42 | #297 |
| 13 | Gemma 4 31B (IT) | 38.37 | #176 |
| 14 | Qwen 3 32B (Non-reasoning) | 38.35 | #443 (Qwen 3 32B) |
| 15 | tiny-aya-water | 38.32 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-italian-summarization-ilpost-sum · How It Works · Data refreshed daily, snapshot 2026-10-20.