text model · Qwen3.6 · iOS
Can I run Qwen3.6 35B-A3B on iPhone 16?
No. Qwen3.6 35B-A3B needs ~24.5 GB even at Q4_K_M, but iPhone 16 only has ~4.5 GB usable.
Needs ~24.5 GB even at Q4_K_M, but only ~4.5 GB is usable.
That figure is at a 4k context and moves about ±15% as context length changes.
The gap is about 20 GB: Qwen3.6 35B-A3B needs roughly 24.5 GB at Q4_K_M and iPhone 16 leaves only about 4.5 GB usable for a model. The lightest tracked hardware that runs Qwen3.6 35B-A3B is the Nvidia GeForce RTX 5090 (32GB) at 32 GB. See Qwen3.6 35B-A3B on Nvidia GeForce RTX 5090 (32GB).
- Q4_K_M needed
- ~24.5 GB
- Usable on device
- ~4.5 GB
- Device memory
- 8 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
- 36B (MoE, 3B active)
- Q4_K_M size
- 22.29 GB
- Q8_0 size
- 36.9 GB
- Context
- 256k
- Ollama tag
- qwen3.6:35b-a3b
- Memory
- 8 GB unified
- Usable for weights
- ~4.5 GB
- Power draw
- ~11 W
- Best runtime
- llama.cpp + Metal (via PocketPal or Off Grid app)
What you can run instead
Run Qwen3.6 35B-A3B on other hardware
FAQ
Can iPhone 16 run Qwen3.6 35B-A3B?
No. Qwen3.6 35B-A3B needs ~24.5 GB even at Q4_K_M, but iPhone 16 only has ~4.5 GB usable.
How much memory does Qwen3.6 35B-A3B need?
iPhone 16 does not have enough memory. At Q4_K_M the weights are ~22.29 GB; with KV cache and runtime overhead, budget ~24.5 GB at a 4k context. It is a Mixture-of-Experts model (36B total / 3B active), so all experts must stay in memory; memory tracks total params, not active params.
What is the best tool to run Qwen3.6 35B-A3B 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/qwen3.6-35b-a3b/iphone-16) Sources
- apple.com
- developer.apple.com
- en.wikipedia.org/wiki/Apple_A18
- en.wikipedia.org/wiki/IPhone_16
- enclaveai.app
- github.com/a-ghorbani
- github.com/google-ai-edge
- github.com/mainframecomputer
- gsmarena.com
- huggingface.co/bartowski
- huggingface.co/Qwen
- huggingface.co/unsloth
- layla-network.ai
- macrumors.com
- ollama.com
- privatellm.app
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.