EuroEval English Summarization - CNN Dailymail: leaderboard
EuroEval's English summarization task on the CNN Dailymail dataset — models write abstractive summaries of English news articles, scored with automatic text-overlap metrics, from the EuroEval European language model benchmark suite.
Metric: chrF3++ (%). Source: euroeval.com. Saturation forecast: Around 2033. 151 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | occiglot-7B-eu5-instruct | 45.54 | |
| 2 | Granite 4.0 Micro | 45.11 | #882 |
| 3 | EuroLLM-9B-Instruct | 45.09 | #1054 |
| 4 | Apertus-70B-Instruct-2509 | 44.93 | #820 |
| 5 | Bielik-11B-v2.3-Instruct | 44.64 | #693 |
| 6 | granite-4.0-h-micro | 43.85 | #1088 |
| 7 | Claude Sonnet 4.5 (Thinking) | 43.73 | #142 (Claude Sonnet 4.5) |
| 8 | Llama 3.1 8B Instruct | 43.69 | #1035 |
| 9 | Llama 3.3 70B Instruct | 43.69 | #534 |
| 10 | Gemini 2.5 Flash Lite (Thinking) | 43.57 | #422 (Gemini 2.5 Flash Lite) |
| 11 | GPT-4.1 | 43.54 | #248 |
| 12 | Qwen 3 14B (Non-reasoning) | 43.31 | #537 (Qwen 3 14B) |
| 13 | Qwen 3 4B (Non-reasoning) | 43.2 | #829 (Qwen 3 4B) |
| 14 | Qwen 3 32B (Non-reasoning) | 43.18 | #443 (Qwen 3 32B) |
| 15 | Claude Haiku 4.5 (20251001) | 43.09 | #261 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-english-summarization-cnn-dailymail · How It Works · Data refreshed daily, snapshot 2026-10-20.