Frequently Asked Questions

Common error codes

• 4xx (client error status codes): Generally indicates request syntax errors, authentication failures, or other issues preventing the request from completing.

• 5xx (server error status codes): Generally indicates server-side errors, such as server crashes or request processing timeouts.

Error code	Possible situations	Solutions
400	Request body format errors, etc.	Check the error content returned by the conversation or the console</a to view the error details and follow the prompts to act. [Common case 1]: If it is a gemini model, you may need to bind a card; [Common case 2]: Data size exceeded the limit, common for vision models — if an image exceeds the upstream single-request traffic limit this error code will be returned; [Common case 3]: Unsupported parameters were added or parameters were filled in incorrectly. Try creating a fresh assistant to test whether it works; [Common case 4]:Context exceeded the limit — clear the context, create a new conversation, or reduce the number of context entries.
401	Authentication failed: model not supported or server account banned, etc.	Contact or check the corresponding service provider account status
403	Request operation not permitted	Follow the error information returned by the conversation orthe consoleerror message prompts to take appropriate action
404	Requested resource not found	Check the request path, etc.
422	Request format is correct, but semantic error	These errors can be parsed by the server but cannot be processed. Common in JSON semantic errors (e.g., null values; a value required to be a string but written as a number or boolean, etc.).
429	Request rate limit reached	Request rate (TPM or RPM) has reached the limit — wait a bit before retrying
500	Internal server error — cannot complete the request	If it persists, contact the upstream service provider
501	The server does not support the requested functionality and cannot complete the request
502	A server acting as a gateway or proxy received an invalid response from the upstream server while attempting to fulfill the request
503	The server is temporarily unable to handle client requests due to overload or maintenance. The delay length may be included in the server's Retry-After header
504	A server acting as a gateway or proxy did not receive a timely response from the upstream server

---

How to view console errors

• Click the Cherry Studio client window and press the shortcut keys Ctrl + Shift + I(Mac: Command + Option + I)

• The currently active window must be the Cherry Studio client window to open the console;

• You must open the console first, then click test or start a conversation or other requests to collect request information.

• In the popped-up console window click Network → Click to view the last entry labeled with a red mark at ② × the completions(when errors occur in conversations, translations, or model connectivity checks) or generations(when errors occur with painting) → ClickResponseto view the full returned content (the area marked ④ in the image).

If you cannot determine the cause of the error, please screenshot this interface and send it to the official group for help.

This inspection method can obtain error information not only during conversations but also during model testing, adding knowledge bases, painting, etc. In all cases you need to open the debug window first, then perform request operations to capture request information.

The name in the Name column (② in the image) varies across different scenarios

Conversation, translation, model checks:completions

Painting:generations

Knowledge base creation:embeddings

---

Formula not rendered / formula rendering error

• If the formula is not rendered and the formula code is shown directly, check whether the formula has delimiters

Delimiter usage

Inline formula

• Use single dollar signs: $formula$

• or use\( and \), for example:\(formula\)

Display formula block

• Use double dollar signs: $$formula$$

• or use \[formula\]

• Example: $$\sum_{i=1}^n x_i$$ $\sum_{i=1}^n x_i$

• Formula rendering errors / garbled text often occur when the formula contains Chinese characters. Try switching the formula engine to KaTeX.

---

Unable to create knowledge base / failed to get embedding dimension

1. Model status unavailable

Confirm whether the provider supports the model or check whether the provider's model service status is normal.

2. Used a non-embedding model

---

Model cannot analyze images / unable to upload or select images

First confirm whether the model supports image recognition. Popular models in Cherry Studio are categorized; models with a small eye icon after their name support image recognition.

Image-capable models will support uploading image files. If the model's function is not correctly matched, find the model in the provider's model list, click the settings button after its name, and check the image option.

Model-specific information can be found on the corresponding provider's site. Like embedding models, models that do not support vision do not need image functionality enabled — checking the image option has no effect.