How fast does Qwen 3.5 4B (GGUF, Vision) run on iPhone?
On an iPhone Air (A19, iOS 26.6.1), Qwen 3.5 4B (GGUF, Vision) generated 14.7 tokens per second in Pocket AI Lab 1.1.0, averaged over 3 runs of 200 tokens (± 2.7).
On the standard llama-bench style probe it read a 512-token prompt at about 158 tok/s and generated 128 tokens at 5.1 tok/s. Cold load took 0.4 s and the process peaked at 1,252 MB. The phone heated up during this run and decode speed dropped 31% between repetitions, so on a cool phone it can be faster.
How much memory and storage does Qwen 3.5 4B (GGUF, Vision) need?
The catalog budgets 4.4 GB of RAM and 3.5 GB of storage for the Q4_K_M build of Qwen 3.5 4B (GGUF, Vision). The measured peak footprint on iPhone Air was 1,252 MB. The 4.4 GB figure is the memory budget the app requires to load the model with its context, while 1,252 MB is what our benchmark run actually used.
It runs on a memory-mapped backend, so the app checks it against the Metal working set rather than the stricter per-app limit. Weights are read from disk as needed, which keeps the measured footprint low.
Which iPhones can run Qwen 3.5 4B (GGUF, Vision)?
Qwen 3.5 4B (GGUF, Vision) runs on iPhones with 8 GB of RAM or more.
| RAM | iPhones | Pocket AI Lab |
|---|---|---|
| 3–4 GB | iPhone 11, 11 Pro, 12, 12 mini, 13, 13 mini, SE (2nd and 3rd gen) | ✗ Too big |
| 6 GB | iPhone 12 Pro, 13 Pro, 14, 14 Plus, 14 Pro, 15, 15 Plus | ✗ Too big |
| 8 GB | iPhone 15 Pro, 16, 16 Plus, 16 Pro, 16e, 17, 17e | ✓ Runs |
| 12 GB | iPhone 17 Pro, 17 Pro Max, iPhone Air | ✓ Runs |
Based on the memory budgets the app measures on each class of iPhone (October 2026). Before downloading, the app checks your exact phone and only offers what fits.
What can Qwen 3.5 4B (GGUF, Vision) do in Pocket AI Lab?
It's a vision-language model, so besides text chat it can describe photos you attach to a message.
How did Qwen 3.5 4B (GGUF, Vision) score on the quality suite?
It passed 81 of 118 tasks, a score of 68.6 out of 100 on suite v3.
Tasks are graded by deterministic checks (exact numbers, regex, word counts, valid JSON), with greedy decoding. See the method.
All runs of Qwen 3.5 4B (GGUF, Vision)
| Model | Gen tok/s | pp512 / tg128 | Load s | Peak MB | Score | Date |
|---|---|---|---|---|---|---|
| Qwen 3.5 4B (GGUF, Vision) | 15.1 ± 1.4 | ~236 / 6.4 | 0.6 | 1,207 | 68.6 | 2026-10-21 |
| Qwen 3.5 4B (GGUF, Vision) | 14.7 ± 2.7 | ~158 / 5.1 | 0.4 | 1,252 | 68.6 | 2026-10-23 |
Other Qwen models in the catalog
- Qwen 2.5 3B, MLX, 4bit, 26.3 tok/s
- Qwen 2.5 14B, MLX, 4bit
- Qwen2-VL 7B (Vision), MLX, 4bit, 11.6 tok/s (throttled)
- Qwen 3.5 0.8B (GGUF, Vision), llama.cpp, Q4_K_M, 73.7 tok/s
- Qwen 3.5 2B (GGUF, Vision), llama.cpp, Q4_K_M, 36.6 tok/s
- Qwen2.5 1.5B (Core ML), Core ML, int8-W8A16, 3.0 tok/s
How do I install Qwen 3.5 4B (GGUF, Vision)?
Open the Models tab in Pocket AI Lab and tap Qwen 3.5 4B (GGUF, Vision). The app checks it against your phone's memory, downloads 3.5 GB from Hugging Face and it's ready for offline chat.