text model · Llama 4 · iOS
Can I run Llama 4 Maverick on iPhone 17 Pro?
No. Llama 4 Maverick needs ~231.7 GB even at Q4_K_M, but iPhone 17 Pro only has ~8 GB usable.
Needs ~231.7 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 223.7 GB: Llama 4 Maverick needs roughly 231.7 GB at Q4_K_M and iPhone 17 Pro leaves only about 8 GB usable for a model. No single tracked device has enough memory; Llama 4 Maverick needs a multi-GPU or high-memory rig.
- Q4_K_M needed
- ~231.7 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
- 400B (MoE, 17B active)
- Q4_K_M size
- 226.09 GB
- Q8_0 size
- 396.57 GB
- Context
- 1000k
- Ollama tag
- llama4:128x17b
- 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
FAQ
Can iPhone 17 Pro run Llama 4 Maverick?
No. Llama 4 Maverick needs ~231.7 GB even at Q4_K_M, but iPhone 17 Pro only has ~8 GB usable.
How much memory does Llama 4 Maverick need?
iPhone 17 Pro does not have enough memory. At Q4_K_M the weights are ~226.09 GB; with KV cache and runtime overhead, budget ~231.7 GB at a 4k context. It is a Mixture-of-Experts model (400B total / 17B active), so all experts must stay in memory; memory tracks total params, not active params.
What is the best tool to run Llama 4 Maverick 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/llama-4-maverick/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.