Cherry Studio with API429: root OpenAI endpoint, URL preview rules and manual model setup

Create a dedicated custom OpenAI provider in Settings

Open Cherry Studio Settings, navigate to Model Provider and click Add Provider. Name the provider API429 so its credentials and model entries stay clearly separated from any built-in vendor presets. Under Endpoint settings, choose the OpenAI protocol type and enter the root host https://gateway.api429.com without a trailing slash or /v1 suffix. Leave secondary protocol fields blank; this setup targets standard OpenAI-compatible Chat Completions.

Enter your API429 client credential in the protected API Key field. In the reference JSON below, YOUR_API429_KEY is an inert placeholder that you replace only inside the application UI. Never store live keys in shared notes, exported JSON snippets or diagnostic screenshots. If multiple people share the same workstation or desktop profile, remember that a saved provider key is accessible to anyone using that local Cherry Studio installation.

Inspect the final endpoint preview and path construction rules

According to the official Cherry Studio provider documentation, entering a standard root host causes the client to append /v1/chat/completions automatically. When you set openai_endpoint to https://gateway.api429.com, the final URL preview displayed beneath the field must read https://gateway.api429.com/v1/chat/completions. Do not paste https://gateway.api429.com/v1 without a trailing slash into Cherry Studio, or the automatic suffix can produce a duplicated /v1/v1/chat/completions path that fails with a routing error.

Trailing punctuation changes how Cherry Studio constructs the request URL: a trailing slash / tells the client to omit the /v1 prefix and append only chat/completions, while a trailing hash # tells the client to use the exact string verbatim without appending anything. The route comparison example below shows all three documented inputs that resolve to https://gateway.api429.com/v1/chat/completions. Use the root origin https://gateway.api429.com as your primary setting and always verify the live preview line in your installed version.

Add gemini-3.8-flash manually and enable the provider toggle

Under the provider Models section, click the + button to add a model manually and enter gemini-3.8-flash as the exact Model ID. You can set Model Name to Gemini 3.8 Flash and Group Name to API429 for readable UI grouping without altering the wire identifier sent in API requests. Add only this single model for your initial test so you can be certain which ID is selected. Manual registration in the UI does not verify account entitlement.

Turn on the provider enable toggle at the top of the settings panel; a disabled provider will not appear in the chat model picker even after its models are saved. Before sending a request, confirm that gemini-3.8-flash appears in the API429 public catalog and verify your key entitlement with an authenticated GET /v1/models request. Do not click test buttons repeatedly if the model is missing from your key allowlist.

Run one bounded text check and treat Model Check as paid

Open a new standard chat conversation with web search, knowledge bases, file attachments and MCP tools turned off, then select gemini-3.8-flash under the API429 provider. Set max_tokens to 256 and turn off streaming if your installed Cherry Studio version exposes those controls for the conversation. Both the field sheet and the target request body below are human reference fixtures rather than importable Cherry Studio backup files.

Note that clicking the built-in Model Check button sends a live request using a conversational model from the provider list (documented as the most recently added model). Treat Model Check and your first manual chat prompt as paid API calls and run only one deliberate check after verifying the final URL preview. Afterwards, compare the response and token accounting in your API429 dashboard rather than assuming one UI click always equals one isolated request.

Separate standard chat from Cherry Agent and total budget

This guide covers only a standard OpenAI-compatible chat provider connection; it does not configure or validate Cherry Agent, which is a separate agent subsystem with its own runtime and protocol requirements. A working text answer also does not prove compatibility for tool calling, vision attachments, embeddings or structured outputs. Finally, setting max_tokens: 256 limits a single reply rather than your overall spend: accumulated prompt history, health checks and additional turns all contribute to total cost. Check current token rates on the model card and pricing comparison pages.

Manual UI field reference sheet; not a Cherry Studio import file

{
  "provider_name": "API429",
  "openai_endpoint": "https://gateway.api429.com",
  "api_key": "YOUR_API429_KEY",
  "model_id": "gemini-3.8-flash",
  "model_name": "Gemini 3.8 Flash",
  "group_name": "API429",
  "enabled": true
}

Three documented Cherry Studio URL route construction rules

[
  {
    "input": "https://gateway.api429.com",
    "final_url": "https://gateway.api429.com/v1/chat/completions"
  },
  {
    "input": "https://gateway.api429.com/v1/",
    "final_url": "https://gateway.api429.com/v1/chat/completions"
  },
  {
    "input": "https://gateway.api429.com/v1/chat/completions#",
    "final_url": "https://gateway.api429.com/v1/chat/completions"
  }
]

Target Chat Completions request contract; not a Cherry Studio preset

{
  "model": "gemini-3.8-flash",
  "messages": [
    {
      "role": "user",
      "content": "Answer in one sentence: what is an API?"
    }
  ],
  "max_tokens": 256,
  "stream": false
}

Payment and a $10 test

• Russian bank card. Top up your balance in the account.
• Invoice for companies. Bank transfer with closing documents.
• USDT and TON. Top up in crypto.

$10 test credit. Subscribe to our Telegram channel through the bot and get $10 on your balance for the first requests. Get $10 in the bot