Interactions API: Submitting a multimodal request routes to model that doesn't support multimodal requests

[gbmerrall]

When submitting a multimodal request with the background parameter the request appears to be routed internally to a model with no multimodal support.

Environment details

• Programming language: Python
• OS: Mac
• Language runtime version: 3.12
• Package version: 2.20.0

Steps to reproduce

1. Craft a multimodal request that includes the background parameter

    uploaded_file = client.files.upload(file="loom.webp")
    
    interaction = client.interactions.create(
        model="gemini-3.7-flash",
        input=[
            {
                "type": "text",
                "text": "What is in this image?",
            },
            {
                "type": "image",
                "uri": uploaded_file.uri,
                "mime_type": uploaded_file.mime_type
            },
        ],
        background=True,
    )

Error text:

Traceback (most recent call last):
  File "/Users/graeme/Code/sandbox/.venv/lib/python3.13/site-packages/google/genai/_gaos/interactions.py", line 711, in create
    return _speakeasy_parse_response(http_res)
  File "/Users/graeme/Code/sandbox/.venv/lib/python3.13/site-packages/google/genai/_gaos/interactions.py", line 489, in _speakeasy_parse_response
    raise errors.CreateInteractionClientError(
        response_data, http_res, http_res_text
    )
google.genai._gaos.errors.createinteraction.CreateInteractionClientError: Image input modality is not enabled for models/gemini-3.7-flash-agent

Making sure to follow these steps will guarantee the quickest resolution possible.

Thanks!

[kkorpal]

Hey @gbmerrall, thanks for bringing this to our attention! I will bring this straight to the engineering team for investigation.