← Back to PocketLLM

Support

Get help with PocketLLM. Find answers below or contact us directly.

Frequently Asked Questions

What devices does PocketLLM support?

PocketLLM runs on iPhone and iPad with iOS 18.0 or later. For best performance, we recommend iPhone 12 Pro or newer (devices with 6 GB+ RAM). Older devices can run smaller models like SmolLM2 (135M).

How much storage do AI models need?

Model sizes range from 80 MB (SmolLM2 135M) to about 4 GB (Llama 3.2 3B). You can download and delete models at any time from Settings.

Does PocketLLM work offline?

Yes. Once you download a model, all AI processing happens on your device. You can chat in airplane mode with no internet connection.

Is my data really private?

Yes. PocketLLM never sends your conversations to any server. All processing is on-device. We don't collect accounts, emails, analytics, or any personal data. Conversations are encrypted with AES-256-GCM and stored only on your device.

How do I upgrade to Pro?

Open PocketLLM, go to Settings, tap your current plan (e.g., "PocketLLM Free"), then tap "Upgrade to Pro." You can choose monthly ($4.99/mo) or annual ($29.99/yr, save 50%). Both include a 1-week free trial.

How do I cancel my subscription?

Subscriptions are managed through the App Store. Go to iPhone Settings > your name > Subscriptions > PocketLLM Pro > Cancel Subscription. You'll keep Pro access until the end of your current billing period.

How do I restore purchases on a new device?

Open PocketLLM, go to Settings, tap your plan, then tap "Restore Purchases." Your Pro subscription will be restored automatically if it's still active.

A model download failed or is stuck. What should I do?

Check your internet connection and available storage. Then go to Settings > Models, delete the partial download, and try again. Models are downloaded from Hugging Face CDN and may take a few minutes on slower connections.

Why are responses slow or lower quality?

Response speed depends on your device and model size. Smaller models (SmolLM2, Qwen 0.5B) are faster but less capable. Larger models (Llama 3.2 1B/3B) produce better responses but need more RAM and processing time. Try a different model if performance doesn't meet your needs.

Contact Us

Email Support

For bug reports, feature requests, or any questions:
[email protected]

We typically respond within 24 hours.