Use your models in other apps
Point coding agents and any OpenAI-compatible tool at your home Mac's models.
Pallm can lend your host’s models to other tools, such as pi, opencode or anything that speaks the OpenAI API, on a laptop away from home. They use the home model as if it ran on the laptop, with no API key from any AI company.
On a client Mac
On a Mac paired as a client, open Settings and switch on Use Pallm’s local model gateway to access your host Mac’s models. Pallm shows two values to copy:
export OPENAI_BASE_URL=http://127.0.0.1:7742/v1
export OPENAI_API_KEY=plk_…Set them in your shell, then start your tool. The gateway works while Pallm is open.
The key belongs to that Mac and only unlocks the models. A program that has it can’t read your chats, delete anything or unpair a device.
From the Terminal, without the app
On any Mac or Linux machine with the pallm command, enrol it once with a pairing link from the host, then connect:
pallm enrol 'pallm://pair?…' # the link from "Pair another device…"
pallm connect # prints the base URL and keyOn the host itself
The host doesn’t need the gateway. Point your tools straight at LM Studio, Ollama or whichever server you run.