Troubleshooting

Error after running an automation action

Expand the global task overlay to view the complete error.

Failed tasks can be retried normally. If a task remains running or queued without making progress for a long time, use Force Retry.

Read the specific error message, then check the troubleshooting sections below.

If you cannot resolve the issue, contact the developers.

Request timeout

This usually indicates network instability or that the provider service is temporarily unavailable. Try:

• checking whether the network connection is stable;
• trying again later;

401 error

The error may look like:

API returned status 401: {"error":{"message":"Authentication Fails, Your api key: ****u2sd is invalid","type":"authentication_error","param":null,"code":"invalid_request_error"}}

This means that the API key is invalid. Check:

1. The API key was copied completely and has no extra spaces.
2. The provider and API URL match.

429 error

This usually indicates provider throttling or excessive concurrency. Try:

• retrying the missing items manually;
• adjusting the rate-limit settings (refer to the provider documentation);
• lowering concurrency;
• increasing the request interval.

Do not immediately select all items for reprocessing. This may make the situation worse by triggering further provider throttling.

400 error

This means that the API request is invalid. There can be many causes, so inspect the error content.

If the error looks like this:

API returned status 400: {"error":{"message":"The supported API model names are deepseek-v4-pro or deepseek-v4-flash, but you passed deepseek-v4-pro1111.","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}

The model identifier is invalid. Check:

1. The identifier was not entered incorrectly. Use the identifier provided by the provider, not the model display name.
2. The provider may have removed the model as obsolete. If so, try another model.

If the error looks like this:

API returned status 400: {"error":{"message":"Failed to deserialize the JSON body into the target type: messages[0]: unknown variant `image_url`, expected `text` at line 1 column 3061412","type":"invalid_request_error","param":null,"code":"invalid_request_error"}}

The model may have been assigned an incorrect capability when the service was configured. For example:

• a non-vision model was assigned a vision-related task.

“Expected X translations but got Y”

For example:

Expected 6 translations but got 7. The AI likely returned malformed content, please
try again.

This means that the model did not provide one translation for each item as required by the system prompt. You can:

• edit the model configuration and increase reasoning effort;
• switch to another model;
• try the same model again later.

If this happens frequently, the model may not be suitable for this task because it does not follow instructions reliably.

The provider-selection page fails to load

Your network may be unavailable. Check whether the device can connect to the Neko Translator main site.

If the network problem cannot be resolved, use Add Manually to configure the provider you want to use.

Text blocks are not detected or source text is not extracted

• Confirm that the source language is set correctly.
• Check that the image is clear and that the text is not too small, tilted, obscured, or blended into the background.
• Try a different detection service.
• Handwriting, sound effects, and decorative fonts can be difficult to recognize. Add a text block manually and enter the source text yourself.

Translate action is disabled

Translate needs at least one text block with source text and a model configured with the Translate capability.

If the source text is empty, run Extract Text or enter the text manually first.

Local model download or import failed

• Check disk space and network access.
• A failed ZIP import or hash verification usually means that the archive is incomplete or has been tampered with.