Core Arc Studio · Free tool

Which LLM can my iPhone run?

Pick your iPhone, iPad or Mac. The checker shows which Qwen, Gemma and Llama models will run well on it, which will be slow or tight on RAM, and which will not fit, with a rough idea of speed for each.

The check · 01 Runs in your browser · Estimates

iPhone 17A19 · 2025

RAM: 8 GB
Usable: 5.6 GB
Run well: 7 of 9

Qwen 3.5 0.8BRuns well
About 1.4 GB in use. Fast. Replies appear quicker than you can read them.
Llama 3.2 1BRuns well
About 1.7 GB in use. Fast. Replies appear quicker than you can read them.
Qwen 3.5 2BRuns well
About 2.1 GB in use. Fast. Replies appear quicker than you can read them.
Gemma 4 2BRuns well
About 2.1 GB in use. Fast. Replies appear quicker than you can read them.
Llama 3.2 3BRuns well
About 2.8 GB in use. Steady. Replies appear at about reading pace.
Qwen 3.5 4BRuns well
About 3.3 GB in use. Steady. Replies appear at about reading pace.
Gemma 4 4BRuns well
About 3.3 GB in use. Steady. Replies appear at about reading pace.
Qwen 3.5 9BWon't fit
Too large for the RAM an iPhone gives one app. Runs on iPad and Mac.
Larger Qwen and Gemma (26B and up)Won't fit
Needs a Mac with 24 GB of RAM or more.

Estimates from RAM and model size, not measurements. Real results vary with iOS version and what else is open.

Run these models in LocalMind AI →

The models in this list are the ones our apps run on your device. LocalMind AI uses them for private offline chat, and DocuMind uses them to answer questions from your own documents.

The working · 02 Rule of thumb

How the estimate is worked out.

RAM the model needs

Parameters times bits per parameter, divided by eight, plus about 1 GB for the conversation and the app. The models here are compressed to about 4.5 bits per parameter, so a 4B model needs a little over 3 GB.

RAM your device allows

iOS never gives one app all of the RAM. The checker assumes about 70% of it, and no more than 6 GB on any iPhone, which is why a 12 GB iPhone does not run everything a 12 GB iPad can. A Mac has no fixed limit, so the checker assumes 75%.

How fast it feels

Speed comes from the chip and the size of the model. A newer chip writes a reply faster, and a model twice the size takes roughly twice as long per word. On a Mac the chip tier matters most: a Max chip runs a large model several times faster than a base chip with the same RAM. The speed notes are rules of thumb, not benchmarks.

The full list · 03 Updated 10 October 2026

Every device, model by model.

The same estimates as the checker, written out for every iPhone, iPad and Mac. Device and model data last updated 10 October 2026.

iPhone+

iPhone 18 Pro / 18 Pro Max (A20 Pro, 12 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 17 Pro / 17 Pro Max (A19 Pro, 12 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone Air (A19 Pro, 12 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 17 (A19, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 16e (A18, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 16 Pro / 16 Pro Max (A18 Pro, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 16 / 16 Plus (A18, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 15 Pro / 15 Pro Max (A17 Pro, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 15 / 15 Plus (A16 Bionic, 6 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B
Tight: Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 14 Pro / 14 Pro Max (A16 Bionic, 6 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B
Tight: Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 14 / 14 Plus (A15 Bionic, 6 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B
Tight: Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone SE (3rd generation) (A15 Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 13 Pro / 13 Pro Max (A15 Bionic, 6 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B
Tight: Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 13 / 13 mini (A15 Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 12 Pro / 12 Pro Max (A14 Bionic, 6 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B
Tight: Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 12 / 12 mini (A14 Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone SE (2nd generation) (A13 Bionic, 3 GB RAM)

Runs well: Qwen 3.5 0.8B
Tight: Llama 3.2 1B
Won't fit: Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 11 Pro / 11 Pro Max (A13 Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone 11 (A13 Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone XS / XS Max (A12 Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPhone XR (A12 Bionic, 3 GB RAM)

Runs well: Qwen 3.5 0.8B
Tight: Llama 3.2 1B
Won't fit: Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad+

iPad Pro (M5), 1 TB or 2 TB (M5, 16 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B
Won't fit: Larger Qwen and Gemma (26B and up)

iPad Pro (M5), 256 GB or 512 GB (M5, 12 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B
Won't fit: Larger Qwen and Gemma (26B and up)

iPad Air (M3) (M3, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad (A16) (A16, 6 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B
Tight: Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad mini (A17 Pro) (A17 Pro, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad Pro (M4), 1 TB or 2 TB (M4, 16 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B
Won't fit: Larger Qwen and Gemma (26B and up)

iPad Pro (M4), 256 GB or 512 GB (M4, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad Air (M2) (M2, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad Pro (M2), 1 TB or 2 TB (M2, 16 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B
Won't fit: Larger Qwen and Gemma (26B and up)

iPad Pro (M2), up to 512 GB (M2, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad Air (5th generation, M1) (M1, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad (10th generation) (A14 Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad Pro (M1), 1 TB or 2 TB (M1, 16 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B
Won't fit: Larger Qwen and Gemma (26B and up)

iPad Pro (M1), up to 512 GB (M1, 8 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad mini (6th generation) (A15 Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad (9th generation) (A13 Bionic, 3 GB RAM)

Runs well: Qwen 3.5 0.8B
Tight: Llama 3.2 1B
Won't fit: Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad Air (4th generation) (A14 Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad Pro (2020) (A12Z Bionic, 6 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B
Tight: Qwen 3.5 4B, Gemma 4 4B
Won't fit: Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad (8th generation) (A12 Bionic, 3 GB RAM)

Runs well: Qwen 3.5 0.8B
Tight: Llama 3.2 1B
Won't fit: Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad Air (3rd generation) (A12 Bionic, 3 GB RAM)

Runs well: Qwen 3.5 0.8B
Tight: Llama 3.2 1B
Won't fit: Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad mini (5th generation) (A12 Bionic, 3 GB RAM)

Runs well: Qwen 3.5 0.8B
Tight: Llama 3.2 1B
Won't fit: Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

iPad Pro (2018), up to 512 GB (A12X Bionic, 4 GB RAM)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B
Tight: Gemma 4 2B
Won't fit: Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Larger Qwen and Gemma (26B and up)

Mac+

Mac with 128 GB RAM or more (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Gemma 4 31B, Qwen 3.6 35B

Mac with 96 GB RAM (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Gemma 4 31B, Qwen 3.6 35B

Mac with 64 GB RAM (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Gemma 4 31B, Qwen 3.6 35B

Mac with 48 GB RAM (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Gemma 4 31B, Qwen 3.6 35B

Mac with 36 GB RAM (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Gemma 4 31B, Qwen 3.6 35B

Mac with 32 GB RAM (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B, Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Gemma 4 31B
Tight: Qwen 3.6 35B

Mac with 24 GB RAM (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B
Tight: Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Qwen 3.6 35B
Won't fit: Gemma 4 31B

Mac with 18 GB RAM (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B
Won't fit: Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Gemma 4 31B, Qwen 3.6 35B

Mac with 16 GB RAM (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B, Qwen 3.5 9B
Won't fit: Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Gemma 4 31B, Qwen 3.6 35B

Mac with 8 GB RAM (Apple silicon)

Runs well: Qwen 3.5 0.8B, Llama 3.2 1B, Qwen 3.5 2B, Gemma 4 2B, Llama 3.2 3B, Qwen 3.5 4B, Gemma 4 4B
Tight: Qwen 3.5 9B
Won't fit: Gemma 4 26B, Qwen 3.5 27B, Qwen 3.6 27B, Gemma 4 31B, Qwen 3.6 35B

Questions · 04 6 answers

Running an LLM on an iPhone.

Short answers to what people ask before trying a local model on a phone.

Can an iPhone run an LLM offline?+

Yes. Any iPhone from the iPhone XR onward can run a small language model with no internet connection. You download the model once, and after that it answers in airplane mode because all of the work happens on the phone. Newer iPhones with 8 GB of RAM or more run models up to about 4 billion parameters comfortably.

How much RAM does a local LLM need?+

A rough rule is a little over half a gigabyte per billion parameters, plus about 1 GB for the conversation and the app. A 1B model needs under 2 GB, a 4B model a little over 3 GB, and a 9B model about 6 GB. An app can only use part of the RAM in the device, so a phone needs noticeably more RAM than the model itself.

Why are local models slow on older iPhones?+

Two reasons. Older chips do less work per second, so every word takes longer to produce. Older iPhones also have 3 or 4 GB of RAM, which leaves little room beside the model, and iOS slows down or closes apps that run short of RAM. Choosing a smaller model is the fix: a 1B model on an iPhone 11 feels better than a 3B model that barely fits.

Is running an LLM on my iPhone private?+

It is as private as an assistant gets. The model runs on the phone itself, so your messages are never sent to a server and nobody else can read or store them. That holds as long as the app you use does not add its own cloud features or tracking. LocalMind AI has no account and sends nothing off the device.

iPhone or Mac: which is better for local models?+

A Mac runs bigger models. iPhones top out at around 4 billion parameters because of the RAM limit iOS puts on apps, while a Mac with 24 GB of RAM or more can run larger versions of the same model families at 26 billion parameters and up. An iPhone is the better choice for quick questions and drafting on the go, and a Mac for long documents and harder reasoning.

How accurate is this checker?+

It is an estimate. It works from the RAM in each device and the size of each model, using the same RAM tiers LocalMind AI uses to decide which models to offer. It does not measure your phone, and real results shift with the iOS version, how long the chat is, and what else is open.