How fast does Phi-3.5 Mini run on iPhone?
On an iPhone Air (A19, iOS 26.6.1), Phi-3.5 Mini generated 20.7 tokens per second in Pocket AI Lab 1.1.0, averaged over 3 runs of 200 tokens (± 2.8).
On the standard llama-bench style probe it read a 512-token prompt at about 123 tok/s and generated 128 tokens at 11.8 tok/s. Cold load took 1.0 s and the process peaked at 3,189 MB. The phone heated up during this run and decode speed dropped 24% between repetitions, so on a cool phone it can be faster.
How much memory and storage does Phi-3.5 Mini need?
The catalog budgets 2.4 GB of RAM and 2.3 GB of storage for the 4bit build of Phi-3.5 Mini. The measured peak footprint on iPhone Air was 3,189 MB.
Its backend copies weights into the app's memory, so the app checks it against the per-app limit iOS enforces. On a 6 GB iPhone that limit is much lower than 6 GB.
Which iPhones can run Phi-3.5 Mini?
Phi-3.5 Mini runs on iPhones with 6 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) | Not measured yet |
| 6 GB | iPhone 12 Pro, 13 Pro, 14, 14 Plus, 14 Pro, 15, 15 Plus | ✓ Runs |
| 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.
Phi-3.5 Mini is an older model. It stays in the catalog for people who already use it, but the app no longer recommends it.
What can Phi-3.5 Mini do in Pocket AI Lab?
It's a text model for chat, drafting, summaries and quick questions, fully offline once downloaded.
How did Phi-3.5 Mini score on the quality suite?
It passed 63 of 118 tasks, a score of 53.4 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 Phi-3.5 Mini
| Model | Gen tok/s | pp512 / tg128 | Load s | Peak MB | Score | Date |
|---|---|---|---|---|---|---|
| Phi-3.5 Mini | 19.9 ± 2.6 | ~134 / 14.2 | 0.7 | 3,192 | 71.7 | 2026-09-03 |
| Phi-3.5 Mini | 23.3 ± 2.7 | ~94 / 14.7 | 1.0 | 3,167 | 71.7 | 2026-09-04 |
| Phi-3.5 Mini | 20.7 ± 2.8 | ~123 / 11.8 | 1.0 | 3,189 | 53.4 | 2026-10-22 |
Other Phi models in the catalog
- Phi-3.5 Mini (GGUF), llama.cpp, Q4_K_M, 23.2 tok/s (throttled)
How do I install Phi-3.5 Mini?
Open the Models tab in Pocket AI Lab and tap Phi-3.5 Mini. The app checks it against your phone's memory, downloads 2.3 GB from Hugging Face and it's ready for offline chat.