Docs · Guide

Empryo on any model

Empryo keeps its settings in ~/.empryo/config.json (a project's .empryo/config.json goes over it). Its providers are a providers array of OpenAI-compatible endpoints that it lists models from and sends /chat/completions to. magpie adds a provider of its own there and sets the model sessions start on to one of magpie's. Empryo then runs on any model in magpie: DeepSeek, GLM, Kimi, Qwen, GPT or a ChatGPT subscription.

Set up Empryo

  1. Add a provider in magpieAdd an API key on the Providers page, or sign in to a subscription.
  2. Pick a model for Empryo on the Agents pageEmpryo's row lists your current Empryo model and the models via magpie. Pick one of the latter.
  3. Restart empryoEmpryo reads its settings at start-up, so reopen sessions that are already open.
Terminal
magpie empryo deepseek/deepseek-v4-pro

On Windows, Empryo in a WSL distro is found too and set up the same way, at a gateway address that distro can reach.

What gets written

Terminal
{
  "providers": [
    …your own providers, left as they are…,
    {"id": "magpie", "name": "Magpie",
     "baseURL": "http://127.0.0.1:3425/v1",
     "modelsAPI": "http://127.0.0.1:3425/v1/models"}
  ],
  "defaultModel": "magpie/deepseek/deepseek-v4-pro"
}

defaultModel is <provider id>/<model id>, split at the first /, so magpie/deepseek/deepseek-v4-pro is deepseek/deepseek-v4-pro under the magpie provider. Your own providers and other settings stay as they are.

magpie checks the wiring: when config.json's magpie provider is gone, or its baseURL isn't the gateway's address any more, Empryo's row says so.

Why there is no envVar

An Empryo provider can name an envVar to read its key from; with none, Empryo sends no key of yours. magpie's provider has none, so you put no key in your environment. With no envVar, Empryo lists a provider only when its baseURL answers (2xx, 401 or 403), which the gateway's GET /v1 does.

When it takes effect

Empryo reads its settings only at start-up. While empryo is running, magpie reminds you to restart open empryo sessions.

Switching back

Switching back puts defaultModel back to what it was and removes the magpie provider. If magpie made config.json and nothing is left in it, the file goes too. You can also pick your own Empryo model again.

Terminal
magpie empryo default

FAQ

How do I use DeepSeek, GLM or a ChatGPT subscription in Empryo?

Pick a via-magpie model for Empryo on magpie's Agents page. magpie adds a provider with id magpie to ~/.empryo/config.json and sets defaultModel to magpie/provider/model. Restart empryo to use it.

Do I need an API key environment variable?

No. magpie's provider has no envVar, so Empryo sends no key of yours; the magpie gateway forwards requests to the providers you added in magpie.

Does a project's .empryo/config.json matter?

Yes. A project's .empryo/config.json goes over ~/.empryo/config.json; if it sets its own defaultModel, that one wins in that project.

Is my old model still there after switching back?

Yes. magpie kept the defaultModel you had and puts it back; your own providers were never changed.

More agents on the same models: Lody on any model, AtomCode on any model, Crush on any model and miniQ on any model. Something fails? See Fixing common errors, or ask on Discord.