1 Install

Go to https://pi.dev/ use curl command

2.

You won't have models,

To use a model in pi you need to visit https://pi.dev/docs/latest/models copy the directory


                        ~/.pi/agent/models.json

paste it into command pallete ^{⌘ + ⇧ + p} to open the file

3.

Very Similiar to everything else,

name it, give it base url (with endpoint v1),

The key you typed when setting up, I did ^'key'

mode name in id

context legnth from lm studio in contextWidow

(make sure model has text, and imgs), and them to array input


{
    "providers": {
        "lmstudio": {
            "baseUrl": "http://127.0.0.1:1234/v1",
            "api": "openai-completions",
            "apiKey": "key",
            "input": ["text","img"],
            "models": [
                {
                    "id": "qwen3.6-35b-a3b"
                }
            ]
        }
    }
}

This didn't load right I had to create a new file in


                            /Users/tt/.pi/agent/models.json

and paste in the above code to for it to work , all done, go work,