iOS app

Know which AI models your phone can run.

WillItRun reads your device's allocatable memory ceiling, not total RAM, and rates every on-device GGUF model in its catalog: runs great, marginal, or will not fit. The verdict arrives before you download a single weight.

Download on the App Store View privacy policy

On-device only. No personal data collected.

WillItRun home screen showing the device memory budget and a See Which Models Run button

Real memory ceiling

iOS keeps a large share of RAM for itself.

WillItRun probes what your app can allocate, the number that decides whether a model loads or crashes.

Fit check

Every catalog model, ranked and rated.

RUNS GREAT, MARGINAL, or WILL NOT FIT for each GGUF model, colour coded against your device's ceiling.

Model fit results listing Gemma 3 1B as RUNS GREAT and several models as MARGINAL

Context matters

Context length changes the verdict.

KV cache grows with the window. Between 4K and 128K tokens, the same model can move from green to amber to out entirely.

Context length slider set to 16K tokens, showing updated model verdicts

Memory breakdown

The full memory budget, itemized.

Resident weights, KV cache, scratch buffers, runtime base, evictable weights, total footprint, survivable ceiling. Plus TTFT and burst speed from your real bandwidth.

Gemma 3 1B detail screen with a memory breakdown: KV cache, weights, TTFT, burst speed

GGUF URL analyzer

Paste a Hugging Face URL.

WillItRun reads 64 KB of metadata over an HTTP range request and rates the model against your device. The weights never download.

Analyze GGUF URL screen with a paste field and an Analyze button

How it works

The verdict, drawn.

WillItRun reads the ceiling iOS will grant, stacks the model's real memory needs against it at your chosen context length, and calls the verdict before you download a single weight.

WillItRun collects no personal data. On-device only.

Read the full privacy policy