Can I run Gemma 4 offline?

Yes. Gemma 4 is a downloadable model: with Your Own AI it runs entirely on your computer, works with no internet connection, and your conversations never leave your device.

What hardware does Gemma 4 need?

The smallest variant (E2B, a 3.1 GB download) runs on a machine with 8 GB of memory. Larger variants want more memory or a graphics card - the interactive checker on this page grades every variant against your hardware, using the same logic as the app.

Is Gemma 4 free to use?

The model weights are released under the Apache 2.0 (Gemma 4 terms) license and download at no charge. Your Own AI itself is free and open source for local use.

Can Gemma 4 see images?

Yes - attach screenshots, charts, or documents and it reads them locally. Nothing is uploaded.

Gemma 4

Runs on your device

Google · Apr 2026

Google's latest open model. Strong writing, analysis, and instruction-following. E2B/E4B run on modest machines; the 26B is a fast MoE.

Context: 128K tokens
Smallest download: 3.1 GB
Runs from: 8 GB RAM
License: Apache 2.0 (Gemma 4 terms)

writinganalysislong contextchatsees imagesMLX for Apple Silicon

Will Gemma 4 run on your machine?

This is the same sizing the app uses - not a marketing estimate.

System memory32 GBGraphics card memory8 GB

E2B3.1 GB downloadRuns fully on your graphics card

E4B5.2 GB downloadRuns fully on your graphics card

12B7.0 GB downloadToo big for this machine

26B-A4B (MoE)10.5 GB downloadRuns - experts in RAM, the rest on the GPU

Mixture-of-experts: the always-on layers sit on your card, the experts sit in system memory - bigger than your card, still quick.

26B-A4B (MoE)14.5 GB downloadRuns - experts in RAM, the rest on the GPU

Mixture-of-experts: the always-on layers sit on your card, the experts sit in system memory - bigger than your card, still quick.

31B17.7 GB downloadToo big for this machine

On Apple Silicon, system memory is also graphics memory - pick "None / integrated / Apple Silicon" and read the bars against your Mac's memory. The app checks your real hardware before every download.

Variants & sizes

Variant	Download	Quantization	Needs	Apple Silicon
E2B	3.1 GB	Q4_K_M	8 GB RAM	MLX · 3.6 GB
E4B	5.2 GB	Q4_0 (QAT)	16 GB RAM	MLX · 6.6 GB
12B	7.0 GB	Q4_0 (QAT)	16 GB RAM	Metal (GGUF)
26B-A4B (MoE)MoE	10.5 GB	UD-Q2_K_XL	24 GB RAM	Metal (GGUF)
26B-A4B (MoE)MoE	14.5 GB	Q4_0 (QAT)	32 GB RAM	MLX · 15.4 GB
31B	17.7 GB	Q4_0 (QAT)	32 GB RAM	Metal (GGUF)

The app picks the right variant for your machine and verifies every download. Model files come straight from their official repositories, pinned to exact revisions.

Context window, compared

Nemotron 3.5 Lightning256K

Ornith 1.5256K

Gemma 4128K

License

Gemma 4 is released under the Apache 2.0 (Gemma 4 terms) - read the terms. The weights download to your machine and stay there; Your Own AI adds no accounts, tracking, or lock-in on top.

Frequently asked questions

Can I run Gemma 4 offline?: Yes. Gemma 4 is a downloadable model: with Your Own AI it runs entirely on your computer, works with no internet connection, and your conversations never leave your device.
What hardware does Gemma 4 need?: The smallest variant (E2B, a 3.1 GB download) runs on a machine with 8 GB of memory. Larger variants want more memory or a graphics card - the interactive checker on this page grades every variant against your hardware, using the same logic as the app.
Is Gemma 4 free to use?: The model weights are released under the Apache 2.0 (Gemma 4 terms) license and download at no charge. Your Own AI itself is free and open source for local use.
How much context does Gemma 4 support?: 128K tokens of context.
Can Gemma 4 see images?: Yes - attach screenshots, charts, or documents and it reads them locally. Nothing is uploaded.

Run Gemma 4 in two clicks

Download Your Own AI, pick Gemma 4, and the app sizes it to your hardware. Private by architecture - no account needed for local use.

Download the app