Troubleshooting
When users encounter issues calling models with API keys, it typically falls into the following four categories. Diagnose based on error codes or messages.
Insufficient Balance
Q1: "Insufficient Balance" or "Quota Exceeded"
Symptoms: API returns 402 Payment Required or 429 Too Many Requests with balance errors.
- Causes:
- Upstream balance exhausted: The enterprise provider account (e.g., Aliyun Cloud Bailian, DeepSeek official) is out of credit.
- Gateway rate limit: Cherry Studio rate limits may trigger short-term throttling.
- Solutions:
- Check in Model Management that the provider API Key is valid.
- Log in to the upstream provider console (e.g., Aliyun Cloud Bailian console) to check account balance.
- Temporary workaround: In Group Management, switch the group's default model to a provider with balance.
Model Not Found
Q2: "Model Not Found"
Symptoms: API returns 404 Not Found with The model 'xxx' does not exist.
- Causes:
- The model name (e.g.,
qwen-plus) is not in the user's group's allowed models.
- The model name (e.g.,
- Solutions:
- Check request body: Verify the
modelparameter in the user's code. - Check group permissions: Go to Group Management > select the user's group > check Model routing to ensure the model is selected.
- Alias mapping: Verify whether model names are remapped (e.g.,
qwen-maxmapped toqwen3-max).
- Check request body: Verify the
Authentication Failure
Q3: "401 Unauthorized"
Symptoms: API rejects the connection.
- Causes:
- Key is incorrect (extra spaces).
- Key is disabled or expired.
- Solutions:
- In Key Settings, ensure Enabled is on.
- Check Expiration time.
- If needed, delete the old key and generate a new
sk-key.
No Healthy Upstream
Q4: "No Healthy Upstream"
Symptoms: API returns 502 Bad Gateway or 503 Service Unavailable.
- Causes:
- All model channels configured for the user's group are offline or timed out.
- Solutions:
- Go to Model Management and click Check to test connectivity.
- Verify the server's network can reach the corresponding provider's API (e.g., dashscope.aliyuncs.com).
Provider Management (Express)
Configure and manage the Express edition provider, offering one-stop aggregated model services and unified quota management.
Default Model Settings
Configure default AI models for the system, including assistant, topic naming, and translation models, balancing user experience and operating costs.