text model · Gemma · iOS
Can I run Gemma 4 26B-A4B on iPhone 17 Pro?
No. Gemma 4 26B-A4B needs ~19 GB even at Q4_K_M, but iPhone 17 Pro only has ~8 GB usable.
Needs ~19 GB even at Q4_K_M, but only ~8 GB is usable.
That figure is at a 4k context and moves about ±15% as context length changes.
The gap is about 11 GB: Gemma 4 26B-A4B needs roughly 19 GB at Q4_K_M and iPhone 17 Pro leaves only about 8 GB usable for a model. The lightest tracked hardware that runs Gemma 4 26B-A4B is the Nvidia GeForce RTX 4090 (24GB) at 24 GB. See Gemma 4 26B-A4B on Nvidia GeForce RTX 4090 (24GB).
- Q4_K_M needed
- ~19 GB
- Usable on device
- ~8 GB
- Device memory
- 12 GB
Which quant fits
How to run it
On iOS use Apple Foundation Models (Built into iOS 26, ~3B on-device model, zero download, fully private.).
- Parameters
- 26.5B (MoE, 3.8B active)
- Q4_K_M size
- 17.04 GB
- Q8_0 size
- 26.86 GB
- Context
- 256k
- Ollama tag
- gemma4:26b-a4b
- Memory
- 12 GB unified
- Usable for weights
- ~8 GB
- Power draw
- ~12 W
- Best runtime
- llama.cpp + Metal (via PocketPal or Off Grid app)
What you can run instead
Run Gemma 4 26B-A4B on other hardware
FAQ
Can iPhone 17 Pro run Gemma 4 26B-A4B?
No. Gemma 4 26B-A4B needs ~19 GB even at Q4_K_M, but iPhone 17 Pro only has ~8 GB usable.
How much memory does Gemma 4 26B-A4B need?
iPhone 17 Pro does not have enough memory. At Q4_K_M the weights are ~17.04 GB; with KV cache and runtime overhead, budget ~19 GB at a 4k context. It is a Mixture-of-Experts model (26.5B total / 3.8B active), so all experts must stay in memory; memory tracks total params, not active params.
What is the best tool to run Gemma 4 26B-A4B on iOS?
On iPhone and iPad, Apple Foundation Models (Built into iOS 26, ~3B on-device model, zero download, fully private.) is the standard choice. Phones realistically run 1B-4B class models. Anything larger thermally throttles or OOMs.
Embed this
A live badge for your model card or README, updated as the data is. [](https://localmodel.run/can-i-run/gemma-4-26b-a4b/iphone-17-pro) Sources
Weights are measured from GGUF files; KV cache and overhead are computed, so totals can vary ~15% with context and runtime. Any tok/s is a bandwidth estimate. See methodology.