gemma-3-27B-pt — benchmark results

Google's pretrained (base) Gemma 3 27B, a multimodal open model with a 128K context window. Provider: Google. Released 2025-03-12. Access: Open.

Unified ELO 1577 ± 9, rank #540 of 1776 rated models, from 395 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EuroEval German Summarization - Mlsum DE41.19Score (%)100
EuroEval Finnish NLU - Scandisent FI93.78Sentiment classification Score (%)99.7
EuroEval Czech NLU - CSFD Sentiment72.91Sentiment classification Score (%)99.6
EuroEval Latvian NLU - MultiWikiQA LV73.01Reading comprehension Score (%)99.6
EuroEval Slovak NLU - CSFD Sentiment SK66.93Sentiment classification Score (%)99.5
EuroEval Swedish NLU - Swerec80.97Sentiment classification Score (%)99.5
EuroEval Ukrainian NLU59.95NLU Average Score (%)99.5
Open Arabic LLM - Arabic MMLU HT Marketing88.03Accuracy (%)99.4
EuroEval Czech NLU61.39NLU Average Score (%)99.1
EuroEval Swedish NLU - Multi Wiki QA SV81.32Reading comprehension Score (%)98.8
EuroEval Greek NLU - MultiWikiQA EL74.2Reading comprehension Score (%)98.5
EuroEval Lithuanian NLU - MultiWikiQA LT74.07Reading comprehension Score (%)98.3

Interactive version: theaggregate.ai/model?slug=gemma-3-27b-pt · How the rankings work · Data refreshed daily, snapshot 2026-07-22.