Skip to content

Vertex AI Search Grounding Metadata not included in Gemini 3.5 response #6512

Description

@tionichm

🔴 Required Information

Please ensure all items in this section are completed to allow for efficient
triaging. Requests without complete information may be rejected / deprioritized.
If an item is not applicable to you - please mark it as N/A

Describe the Bug:
As Gemini 2.5 is getting deprecated, we are forced to move to Gemini 3.5. Our application strictly requires that grounding metadata are included in the model response. We use VertexAiSearchTool (data store that indexes text files in a bucket). We infer the URL of the internal documentation from the object URI. Furthermore, we have safeguards in place that reject a response that has no grounding chunks.

Steps to Reproduce:
Please provide a numbered list of steps to reproduce the behavior:

  1. Use VertexAISearchTool.
  2. Use Gemini 3.5 Flash in the EU region.
  3. Attempt to extract grounding metadata during an after_model_callback.

Expected Behavior:
A clear and concise description of what you expected to happen.

Observed Behavior:
What actually happened? Include error messages or crash stack traces here.

Environment Details:

  • ADK Library Version: 2.5
  • Desktop OS: macOS + Linux
  • Python Version (python -V): >=3.13

Model Information:

  • Are you using LiteLLM: No
  • Which model is being used: gemini-3.5-flash, gemini-3.5-flash-lite

🟡 Optional Information

Providing this information greatly speeds up the resolution process.

Regression:
N/A

Logs:
N/A

Screenshots / Video:
N/A

Additional Context:
N/A

Minimal Reproduction Code:
Instantiate an agent with the appropriate tool

How often has this issue occurred?:

  • Always (100%)

Metadata

Metadata

Assignees

Labels

models[Component] This issue is related to model support

Type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions