No answer content returned for models

Gemini API
, ,
1

Dear Support Team,

I am encountering a consistent issue when using the following two models via the API:

  • deep-research-max-preview-04-2026

  • deep-research-preview-04-2026

Expected behavior:
The API should return a valid, non-empty answer field containing the research output.

Actual behavior:
The response returns a "status": "completed" but the "answer" field only contains a source placeholder, not the actual researched content.
No data appears in references, progress remains null, and error is also null.
The response format is as follows:
[GEMINI RESPONSE] {
“status”: “completed”,
“answer”: “\n\nSources:\n1. dacrbf7qdtba (fileSearchStores/course4986960bea-5wmx0rj1ejuk)\n”,
“references”: ,
“progress”: null,
“error”: null
}

Impact:
The feature is effectively non-functional for these two specific preview models, as no useful answer text is returned.Additional context:
Other deep-research models (e.g., earlier versions) work as expected. The issue is reproducible 100% of the time with the above model names.

Could you please:

  1. Confirm whether these model names are currently active and correctly implemented.

  2. Investigate why the answer field is not populated with the research result.

  3. Advise if any additional parameters or changes in request format are required for these specific preview models.

Thank you for your assistance.

2

I’m getting similar issues in gemini-2.5-flash, from today gemini doesn’t output any outputs or tool calls in over 60% of the occasion