EuroEval Portuguese Summarization - Publico: leaderboard
EuroEval's Portuguese summarization task on the Publico dataset — models write abstractive summaries of Portuguese 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 | Apertus-70B-Instruct-2509 | 34.74 | #820 |
| 2 | Qwen 2.5 72B Instruct | 34.58 | #454 |
| 3 | Qwen 3 30B A3B 2507 Instruct | 34.58 | #488 |
| 4 | GPT-5.2 | 34.41 | #110 |
| 5 | EuroLLM-9B-Instruct | 34.39 | #1054 |
| 6 | GPT-4.1 | 34.38 | #248 |
| 7 | Ministral-3-8B-Instruct-2512 | 33.81 | #661 |
| 8 | Claude Sonnet 4.5 (Thinking) | 33.74 | #142 (Claude Sonnet 4.5) |
| 9 | Gemini 3 Flash (Preview) (Non-reasoning) | 33.57 | #82 (Gemini 3 Flash (Preview)) |
| 10 | Ministral-3-14B-Instruct-2512 | 33.56 | #615 |
| 11 | Claude Sonnet 4.6 | 33.44 | #90 |
| 12 | Claude Haiku 4.5 (20251001) | 33.44 | #261 |
| 13 | Claude Sonnet 4.5 (Non-reasoning) | 33.4 | #142 (Claude Sonnet 4.5) |
| 14 | GPT-5 Mini (2025-08-07) | 33.37 | #172 |
| 15 | Ministral-3-3B-Reasoning-2512 | 33.35 | #932 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-portuguese-summarization-publico · How It Works · Data refreshed daily, snapshot 2026-10-20.