Support

1. Install Ollama from ollama.com and download a model, for example ollama pull gemma3.
2. Ollama only listens to its own computer by default. Quit it and start it so other devices can reach it: OLLAMA_HOST=0.0.0.0 ollama serve.
3. Find your Mac’s address under System Settings › Wi-Fi › Details (for example 192.168.1.20).
4. In ScanLM, open Settings › Local LLM, add a server, choose Ollama, enter the address (port 11434) and pick the model.
5. Allow access when iOS asks about your local network.

1. In LM Studio, load a model and open the server (Developer) tab.
2. Start the server and switch on Serve on Local Network.
3. In ScanLM, open Settings › Local LLM, add a server, choose LM Studio, and enter your computer’s address with port 1234.

llama.cpp’s server and any other OpenAI-compatible server work the same way; choose the matching type and port.

Install a private network such as Tailscale on both your computer and your iPhone, and enter the computer’s Tailscale address in ScanLM. The connection stays between your own devices.

Any current instruction-tuned model works for names, tags, summaries and chat; small models (a few billion parameters) are fast and usually good enough. For pages with no recognisable text, such as handwriting or photos, use a model that understands images and switch on Send page images for that server.

• Check that the iPhone and the computer are on the same network (or both on Tailscale).
• Make sure the server listens on the network, not only on its own computer (see the Ollama and LM Studio steps above).
• Allow ScanLM under iOS Settings › Privacy & Security › Local Network.
• A firewall on the computer may block the port; allow incoming connections for Ollama or LM Studio.

It is listed in Settings › Local LLM. Apple’s on-device model needs an iPhone or iPad that supports Apple Intelligence, iOS 26 or later, and Apple Intelligence switched on in iOS Settings. The model may need a while to download after you turn it on. It works with recognised text only, not with page images.

Inside the app, and in the Files app under On My iPhone › ScanLM. If iCloud Drive upload is switched on, a finished PDF (or images) of each document also appears in iCloud Drive › ScanLM, so you can open it on any device.

Choose the languages you scan in ScanLM’s Settings › Text Recognition › Languages, and put the most common one first. Good light, a flat page and the black-and-white text preset give the best results.

Yes. ScanLM asks iOS to keep a request running in the background; on iOS 26 and later its progress also appears in the system. iOS can still end it if the device needs the resources; the app then tells you and you can try again.

Only where you tell it to: your own model server, your iCloud Drive, or whatever you pick when sharing. Nothing goes to the developer. See the privacy policy.