Skip to main content
Ollama ExplorerBeta
Generalintermediate

gemma3n

Google DeepMindGemma

Gemma 3n models are designed for efficient execution on everyday devices such as laptops, tablets or phones.

1.2M pullsUpdated Jun 26, 20259 tags32K context

Quick start

ollama run gemma3n

Available sizes

TagSizeQuantizationContextMin RAM
gemma3n:e2b5.6GBq4_k_m32K context7 GB
gemma3n:latest7.5GBq4_k_m32K context9.4 GB

Strengths & Limitations

Strengths

  • Efficient execution
  • Designed for everyday devices
  • Good for on-device tasks

Benchmarks

BenchmarkScoreUnit
HellaSwag10Accuracy
PIQA0Accuracy
ARC-c25Accuracy
ARC-e0Accuracy
WinoGrande5Accuracy
MMLU (ProX)0Accuracy
MMLU0Accuracy

Related models