EuroEval Portuguese NLU - HAREM — leaderboard

Metric: Named entity recognition Score (%). Source: euroeval.com. 438 models tracked.

Top models

#ModelScore
1Gemma 4 31B (IT)63.85
2Gemma 4 26B A4B (IT)63.12
3Qwen 3 4B 2507 (Thinking)62.2
4Claude Sonnet 4.5 (Thinking)61.29
5Llama 3.1 70B61.1
6Qwen 3 Next 80B A3B (Thinking)60.81
7Mistral Small 3.160.73
8OLMo 3.1 32B (Thinking)60.41
9Llama 3 70B59.36
10Mistral Small 3.259.09
11QwQ-32B58.1
12O3 Mini (2025-01-31)57.18
13Magistral Small56.9
14Llama 3 8B Instruct56.38
15Qwen 3 30B A3B56.08

Interactive version: theaggregate.ai/benchmark?slug=euroeval-portuguese-nlu-harem · How the rankings work · Data refreshed daily, snapshot 2026-07-22.