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.
Aggregate Rating 1582 ± 17, rank #502 of 1656 rated models, from 471 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval German Summarization - Mlsum DE | 41.19 | chrF3++ (%) | 100 |
| Swallow - Pre-trained English - XWINO | 92.3 | Accuracy (%) | 100 |
| EuroEval Finnish NLU - Scandisent FI | 93.78 | Matthews correlation coefficient (%) | 99.7 |
| EuroEval Czech NLU - CSFD Sentiment | 72.91 | Matthews correlation coefficient (%) | 99.6 |
| EuroEval Latvian NLU - MultiWikiQA LV | 73.01 | Character F1 (%) | 99.6 |
| EuroEval Slovak NLU - CSFD Sentiment SK | 66.93 | Matthews correlation coefficient (%) | 99.5 |
| EuroEval Swedish NLU - Swerec | 80.97 | Matthews correlation coefficient (%) | 99.5 |
| EuroEval Ukrainian NLU | 59.95 | NLU Average Score (%) | 99.5 |
| Open Arabic LLM - Arabic MMLU HT Marketing | 88.03 | Accuracy (%) | 99.4 |
| EuroEval Czech NLU | 61.39 | NLU Average Score (%) | 99.1 |
| EuroEval Swedish NLU - Multi Wiki QA SV | 81.32 | Character F1 (%) | 98.8 |
| EuroEval Greek NLU - MultiWikiQA EL | 74.2 | Character F1 (%) | 98.5 |
Interactive version: theaggregate.ai/model?slug=gemma-3-27b-pt · How It Works · Data refreshed daily, snapshot 2026-10-20.