Gemma 3n E4B — benchmark results

Google's effective-4B on-device Gemma 3n model (June 2025) taking image, audio, video, and text within roughly 3GB of memory. Provider: Google. Released 2025-06-26. Access: Open.

Unified ELO 1473 ± 10, rank #897 of 1776 rated models, from 119 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EuroEval Finnish NLU - Scandisent FI93.19Sentiment classification Score (%)97.6
EuroEval German NLU - Germanquad68.37Reading comprehension Score (%)95.9
EuroEval Finnish NLU - Tydiqa FI75.84Reading comprehension Score (%)94.6
EuroEval Icelandic NLU - NQII60.23Reading comprehension Score (%)94.5
EuroEval Portuguese NLU - MultiWikiQA PT77.48Reading comprehension Score (%)94.5
EuroEval French NLU - FQuAD74.51Reading comprehension Score (%)93.6
EuroEval Norwegian NLU - Norquad74.34Reading comprehension Score (%)93.5
EuroEval Swedish NLU - Swerec79.51Sentiment classification Score (%)93.4
EuroEval Spanish NLU - MLQA ES67.25Reading comprehension Score (%)89.7
EuroEval Faroese NLU - FoQA74.21Reading comprehension Score (%)86
EuroEval Dutch NLU - SQuAD NL77.8Reading comprehension Score (%)85.5
EuroEval English NLU - SQuAD84.71Reading comprehension Score (%)85

Interactive version: theaggregate.ai/model?slug=gemma-3n-e4b · How the rankings work · Data refreshed daily, snapshot 2026-07-22.