Summary
API calls using image generation models (Nano Banana 2, Nano Banana Pro) do not appear in the Logs and Datasets section of Google AI Studio, even though text generation calls from the same API key and
project are logged correctly.
Environment
- Project: Billing-enabled Google Cloud project
- Logging: “Enable logging” is toggled ON
- Models used:
- Text generation (e.g., Gemini 3.1 Pro Preview) → Logs appear normally (~11,000 logs)
- Image generation (Nano Banana 2 / Nano Banana Pro) → No logs found (0 of 0)
Steps to Reproduce
- Enable logging in Google AI Studio → Logs and Datasets
- Make a GenerateContent API call with a text-only model → Log appears
- Make a GenerateContent API call with an image generation model (e.g., gemini-3.1-flash-image-preview) using responseModalities: [“IMAGE”, “TEXT”] → No log appears
- Filter by “Nano Banana 2” or “Nano Banana Pro” in the model dropdown → “No logs found”
Expected Behavior
Since Nano Banana models use the standard GenerateContent / StreamGenerateContent API (not Imagen or Veo), and the https://ai.google.dev/gemini-api/docs/logs-datasets states that “All GenerateContent and
StreamGenerateContent API calls are supported,” these calls should be logged.
Actual Behavior
Only text generation calls are logged. Image generation calls using the same API key and project are completely absent from Logs and Datasets.
Notes
- The documentation excludes Imagen and Veo models from logging, but Nano Banana models are not Imagen — they use GenerateContent.
- The model names (Nano Banana 2, Nano Banana Pro) do appear in the Logs page model filter dropdown, suggesting they should be supported.
- This may be related to Missing Live models from Logs and Datasets in AI Studio , where other GenerateContent-based models are also not logged.
Is this a known limitation, or is it a bug? Any timeline for supporting image generation model logging?