[BUG] GeminiModel failure when using Vertex AI mode without tools

[pshiko]

Checks

• I have updated to the lastest minor and patch version of Strands
• I have checked the documentation and this is not expected behavior
• I have searched ./issues and there are no duplicates of my issue

Strands Version

1.12.0

Python Version

3.12.11

Operating System

macOS 15.6

Installation Method

other

Steps to Reproduce

1. Set the environment variable to enable Vertex AI mode:

   import os
   os.environ["GOOGLE_GENAI_USE_VERTEXAI"] = "True"

2. Create an agent without tools:

   from strands import Agent
   from strands.models.gemini import GeminiModel
   
   # This fails with 400 INVALID_ARGUMENT
   Agent(model=GeminiModel(model_id="gemini-2.5-flash"))("who are you?")

3. Observe the error
   Note: Creating an agent WITH tools works correctly:

   from strands_tools import calculator
   Agent(model=GeminiModel(model_id="gemini-2.5-flash"), tools=[calculator])("who are you?")  # Works fine

Expected Behavior

Agents without tools should work correctly in both Vertex AI mode and standard Gemini API mode.

Actual Behavior

When using Vertex AI mode, the agent fails with the following error:

google.genai.errors.ClientError: 400 INVALID_ARGUMENT.
{'error': {'code': 400, 'message': "The GenerateContentRequest proto is invalid:\n
* tools[0].tool_type: required one_of 'tool_type' must have one initialized field",
'status': 'INVALID_ARGUMENT'}}

• ✅ Works correctly with standard Gemini API (using API key)
• ❌ Fails in Vertex AI mode (GOOGLE_GENAI_USE_VERTEXAI=True)

Additional Context

Environment:

• Model: gemini-2.5-flash (and other Gemini models)
• Deployment mode: Vertex AI (GOOGLE_GENAI_USE_VERTEXAI=True)

Impact:

• 6 integration tests failing in test_model_gemini.py when run in Vertex AI mode
• Users cannot create agents without tools when using Vertex AI

Root Cause:
The current implementation always sends an empty tools array when no tools are provided. Vertex AI's backend strictly enforces the oneof proto definition for FunctionDeclarations, which does not accept empty arrays. The standard Gemini API is more lenient and accepts empty arrays.

Possible Solution

Modify _format_request_config in src/strands/models/gemini.py to only include the tools field when tools are actually provided:

tools=self._format_request_tools(tool_specs) if tool_specs else None,

This approach omits the tools field entirely when no tools are provided, which is accepted by both Vertex AI and the standard Gemini API.

Related Issues

No response