Use ZhiFlo in Cherry Studio

This path is for people who do not want to touch the command line. At the end, the ZhiFlo model appears on the chat screen and returns a real reply.

Before you start

• Install and open Cherry Studio using its official documentation.
• Create a Token dedicated to this device.
• Copy a complete model ID available to your account.

1. Add a custom provider

1. Open Settings.

2. Go to Model Services.

3. Select + to add a custom provider.

4. Choose OpenAI as the provider type.

5. Name the provider ZhiFlo.

6. Enter the connection values:

   Field	Value
   API address / Base URL	https://api.zhiflo.com/v1
   API Key	The ZhiFlo Token for this device
   Model ID	The complete model ID from your account

   Paste the Token only into the credential field. Keep it out of provider names, model labels, and screenshots.

7. Save the provider.

8. Turn on the provider’s enable switch.

Saving and enabling are separate actions. Once enabled, ZhiFlo should appear among the available model services.

2. Check the fields, then send a real message

If the provider page has a Check action, you can run it first. A passing check covers only part of the address or credential setup; it does not confirm that a model returned content.

1. Return to the chat screen.

2. Select the ZhiFlo model from the model picker.

3. Send:

   Reply with OK only.

The setup is complete only when OK appears in the conversation. A green check or a saved provider name is not enough.

Quick recovery

• No ZhiFlo model in the picker: return to Model Services, confirm the provider is saved and enabled, and check that the model ID was added.
• The check passes, but chat fails: verify that the chat screen currently uses the ZhiFlo model, then compare the exact model ID. A credential check is not a model call.
• 401 response: paste an active Token again and remove surrounding whitespace.
• 404 response: use https://api.zhiflo.com/v1 as the API address; do not append /chat/completions to the Base URL.

Next: Understand usage.