Error in batch image generation api using gemini-2.5-flash

Gemini API
,
1

I am trying to do a batch image generation using the Gemini 2.5 flash image (nano banana), and I keep getting errors for the request:

{'error': {'code': 3, 'message': 'Request contains an invalid argument.'}, 'key': *"Modified by Moderator"*

The batch api accepts the data that I send, it shows me that the generation has succeeded, but it then throws this error given above.

The format that I’m using to send the image references in the prompt is:

{
  "key": ".....",
  "request": {
    "contents": [
      {
        "parts": [
          {
            "text": "Generate an image of this pet sitting still in front of the camera. With the head and shoulders of the pet within the frame. The pet should look happy with a party hat on its head. Keep the background colour a uniform colour with colour code #00B140. Only keep the head and the shoulder of the pet in the image. Make sure that the pet fills in the frame as much as possible"
          },
          {
            "file_data": {
              "file_uri": "files/t85xlkojz2x4",
              "mime_type": "image/png"
            }
          }
        ]
      }
    ]
  }
}

I cannot find the official documentation on Google for passing images in the prompt for batch endpoints.

I’m using Python 3.10 on Linux.

2

Hey! Just flagged this to our Batch API team, hang tight!

4

Hi ahar,

Thanks for reporting this issue. The format you are using to send images is not correct. Please try to use a fully qualified URI instead of the file ID.

For example, replace:

files/t85xlkojz2x4

for

https://generativelanguage.googleapis.com/files/t85xlkojz2x4

in the request you shared.

We are improving the error message to make this issue easier to debug.

1 Like