Connect your desktop client.
Current Cherry Studio documentation includes a built-in New API provider. Connect it to PrismRoute, then select the protocol supported by your model.
On this page
Configure the provider
- Open Settings → Model Provider and select the built-in New API provider. Labels can vary with the client language and version.
- Enter your PrismRoute API key and the root API address below. If your version supports renaming, use PrismRoute as the display name.
- Choose the OpenAI Chat protocol for the Chat Completions examples in this guide. Other protocols require matching support from the selected model and account.
| Field | Value |
|---|---|
| Provider | New API; display name PrismRoute where available |
| API Key | Your PrismRoute API key |
| API address / host | https://codex.aicpr.icu |
| Protocol for this guide | OpenAI Chat / Chat Completions |
| Model ID | The exact Chat Completions model ID from your catalog |
Use the host in Cherry Studio
For this provider, Cherry Studio appends the protocol version path itself. Enter https://codex.aicpr.icu without /v1 or /chat/completions. The /v1 base URL in our Python and JavaScript guides is an SDK setting; this client field takes the root host.
https://codex.aicpr.icuhttps://codex.aicpr.icu/v1/chat/completionsIf your installed version does not offer New API, follow its custom OpenAI-compatible provider settings and the official provider guide below. Confirm the final URL above before sending a message. An existing client setup does not need to be upgraded just to follow the SDK guides.
Add and select a model
- Open the PrismRoute model catalog while signed in. Copy a model ID that your key can access and that supports Chat Completions.
- In Cherry Studio, fetch the model list and add the model, or add its exact ID manually. A display name can differ; the model ID must match.
- When ready to send a real model request, use the provider’s Check action. Enable the provider using its top-right switch, then select its model in a chat.
The Check action sends a model request. Fetching a model list and generating a message are separate operations. A listed model does not establish tool, vision, Responses or other protocol support. Start with a short text message before adding optional features.
If the connection fails
| Symptom | Next check |
|---|---|
| 401 or invalid key | Check the API Key field and remove accidental spaces. |
| 404 or unknown endpoint | Inspect the final URL and the selected protocol. |
| Model not found or denied | Check the exact model ID, key restrictions and account group. |
| Requests go to an old address | Inspect any protocol-specific address overrides left in the provider settings. |
| Text works but tools fail | Check the selected model and endpoint’s feature support. |