← All models

Gemma 3N

Google · 2025 · Efficient multimodal · edge

Gemma 3n is tuned for phones, tablets, and laptops. Effective 2B and 4B footprints keep quality high without a workstation, so Delta stays fast on battery-powered machines.

Maker
Google
Architecture
Efficient multimodal · edge
Fits from
Laptop · 7 GB+ RAM
Context
up to 33K
Runs
Local Intelligence
VisionEdgeMultilingualPhones and tabletsTravel laptopsEveryday chat

Sizes

Gemma 3N E2BQ4_K_M

requires 7 GB+ memory · 33K context · Laptop

3 GBWeights
Gemma 3N E4BQ4_K_M

requires 9 GB+ memory · 33K context · Laptop

4.5 GBWeights

Memory at context length

Approximate working set for Gemma 3N E2B as context grows.

4K
3.8 GB
8K
4 GB
16K
4.2 GB
33K
4.5 GB