#6533 Error during subtitle translation stage [Gemini AI] Translation returned empty output

This error means the Gemini API returned no translated text at all. It is usually caused by an invalid/expired API key, no proxy or an unreachable Gemini endpoint, an incorrect model selection, or the “Send full subtitles” mode causing the model to return an empty response.

1. Check your Gemini API configuration:

   • Go to Menu → Translation Settings → Gemini AI.
   • Verify the API Key is correct, has not expired, and has available quota.

2. Verify the network connection:

   • Gemini is a foreign service. If you are in mainland China, you must configure a valid proxy in the main interface (Network proxy field), e.g. http://127.0.0.1:7890.
   • Clear any invalid proxy settings, otherwise all requests will fail.

3. Change the subtitle sending mode:

   • Uncheck Send complete subtitle (i.e. turn off SendSRT). This forces line-by-line translation and often avoids empty responses caused by context length or prompt issues.
   • If you keep it enabled, reduce the AI translation batch subtitle line count in Menu → Tools/Options → Advanced Options → Subtitle Translation Adjustment.

4. Confirm the model:

   • Select a proper Gemini text model from the dropdown (e.g. gemini-2.5-flash). Do not manually enter a non-text model.
   • Make sure you did not accidentally choose a speech recognition or ASR model.

5. Try another translation channel to isolate the problem:

   • Switch to DeepSeek or OpenAI temporarily. If they work, the issue is specific to your Gemini configuration.
6. If the problem continues, check the latest log file in logs/ folder and look for preceding errors such as 401, 403, 429, or token length exceeded. Provide the last 30 lines of the log (and your exact version number, if not already shown) for further diagnosis.

Please refer to the related documentation:

• Gemini Translation
• Common Network Connection Errors