Gemini 3.7 Flash throws "Invalid argument" error when processing UI design images for Sipnena platform

Google AI Studio
1

I am encountering a persistent bug while working on the UI redesign for Sipnena, an AI-powered e-learning platform. Whenever I upload a comprehensive UI design kit containing buttons, icons, and layout tokens and ask the model to help translate elements into front-end code, the system immediately fails.

Environment & Details

  • Model: Gemini 3.7 Flash

  • Platform: Google AI Studio

  • Error Message: Invalid argument (accompanied by: There was an unexpected error. Finish what you were doing.)

Steps to Reproduce

  1. Open a Google AI Studio prompt using the Gemini 3.7 Flash model.

  2. Upload a detailed PNG design board containing application components (such as buttons, cards, progress bars, and icon sets).

  3. Provide a text prompt asking the model to build or restructure layouts using those components.

  4. Click Run.

Expected Behavior

The model should successfully parse the visual layout tokens and output the requested code structures or component breakdowns.

Actual Behavior

image (4)
image (4)1919×1011 205 KB

The execution halts instantly with an “Invalid argument” error before any response or code generation can begin. Has anyone else experienced this multimodal parsing issue, or is there a workaround for handling detailed UI boards in AI Studio?

2

AI Studio is completely unusable on my account. Image generation fails across multiple models, including Nano Banana 2 and Nano Banana Pro. The problem persists after creating a new chat/project and is not limited to a single model. I have an active Tier 1 Postpay billing project, but AI Studio has also created a new Default Gemini Project showing Free Tier. Please investigate whether my account/project has been incorrectly restricted or affected by a backend billing/permission synchronization issue.