XylemNode

Integration guides

Cherry Studio

A cross-platform, open-source desktop AI client (Windows / macOS / Linux) that gathers many models in one interface. Add a custom provider of the OpenAI type to bring XylemNode in.

Get API Key
Site / download
  1. 01

    Install

    Download the installer for your OS from Cherry Studio's official repo / site, install and open it.

  2. 02

    Create an OpenAI-type provider

    Go to Settings → Model Services, click 'Add' at the bottom of the list, give it a label (e.g. XylemNode), choose provider type OpenAI, and confirm.

  3. 03

    Enter the key and API address

    Enter the key and API address. This differs from other tools: enter only the domain for the API address — Cherry Studio appends /v1/chat/completions automatically, so don't add /v1 yourself or it will double up.

    API key
    sk-your-key
    API address
    https://api.xylemnode.com
  4. 04

    Add models and enable

    Click 'Manage' at the bottom-left to auto-fetch models and add them with + (or type a model name by hand, e.g. gpt-5.6-sol). Use 'Check' next to the key box to test, then flip the switch in the top-right to enable this provider.

About the API address

1) Cherry Studio appends /v1/chat/completions after the address you enter, so normally just enter the root https://api.xylemnode.com and don't add /v1 yourself. 2) To pin a full address and stop the auto-append, enter https://api.xylemnode.com/v1/chat/completions ending with # — it then uses only what you typed.

Common questions

Cherry Studio returns 404, or the path looks duplicated

This one is the opposite of the other tools: enter only the domain and let it append /v1/chat/completions itself — its documentation says plainly that it completes the rest of the path and that an address which does not follow that form may simply not work. Adding /v1 by hand is what produces the duplicated path. The docs also mention ending the address with # to pin a full URL, but that does not hold on every version (the repository has issues where a path was still appended after the #), so if you hit a duplicated path, going back to just the domain is the reliable route.