litellm - 💡(How to fix) Fix [Bug]: Gemini 3.5 Flash via Vertex AI reasoning effort broken

Check for existing issues

• I have searched the existing issues and checked that my issue is not a duplicate.

What happened?

When running the proxy using Gemini 3.5 Flash via Vertex AI and specifying the reasoning effort, all requests fail with an error. The bug happens on v1.86.2 but not on 1.85.1.

Official docs saying Gemini 3.5 Flash via Vertex AI supports multiple thinking levels: https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/thinking

Proxy config:

litellm_settings:
  vertex_location: global
  vertex_project: ...

model_list:
  - model_name: gemini-3.5-flash
    litellm_params:
      model: vertex_ai/gemini-3.5-flash
      reasoning_effort: high
      web_search_options:
        search_context_size: high

Steps to Reproduce

1. Set up proxy with Gemini 3.5 Flash via Vertex AI with a reasoning effort
2. In LiteLLM Playground, select the model and the /v1/chat/completions endpoint.
3. Send ping as the message.

Relevant log output

Error fetching response:Error: 400 litellm.UnsupportedParamsError: vertex_ai does not support parameters: ['reasoning_effort'], for model=gemini-3.5-flash. To drop these, set litellm.drop_params=True or for proxy:

litellm_settings:  drop_params: true
.
If you want to use these params dynamically send allowed_openai_params=['reasoning_effort'] in your request.. Received Model Group=gemini-3.5-flash
Available Model Group Fallbacks=None

What part of LiteLLM is this about?

Proxy

What LiteLLM version are you on ?

v1.86.2

Twitter / LinkedIn details

No response