How to use grounding with ChatVertexAI

[jiwonmik]

Checked other resources

• I added a very descriptive title to this question.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.

Commit to Help

• I commit to help with one of those options 👆

Example Code

from google.cloud.aiplatform_v1beta1.types import Tool as VertexTool

llm_class = ChatVertexAI()
tools = [VertexTool(google_search={})]

...
response = llm_class.invoke(prompt_value, tools=self.tools)

Description

I'm trying to use Google search grounding from VertexTool with ChatVertexAI.

What I've understood so far is that

1. I can use grounding google search only with ChatVertexAI (Reference fromlangchain-ai)
2. Above code - using grounding google search is available only for gemini-1.0, gemini-1.5, and gemini-2.0-flash-exp models

The documentation says grounding is available for gemini-flash-2.0 model, I've tried with it with it but the llm's response didn't included grounding_metadata, which implies it didn't triggered grounding - google search.

1. Will gemini-2.0-flash be able to use grounding google search soon?
2. Is this correct - we can use google search grounding only with ChatVertexAI, not ChatGoogleGenerativeAI?

System Info

langchain==0.3.14
langchain-anthropic==0.3.1
langchain-community==0.3.14
langchain-core==0.3.31
langchain-google-community==2.0.4
langchain-google-genai==2.0.4
langchain-google-vertexai==2.0.13
langchain-groq==0.2.1
langchain-openai==0.3.2
langchain-text-splitters==0.3.5
langcodes==3.4.1
langgraph==0.2.45
langgraph-checkpoint==2.0.10
langgraph-checkpoint-postgres==2.0.10
langgraph-sdk==0.1.35
langsmith==0.1.141
language_data==1.2.0