### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

I am sending a chat completion request to the LiteLLM proxy using a model that supports reasoning (specifically `gemma-4-31b-it`). Even when explicitly setting `"think": false` (or `"false"`) in the request body, the response still includes `reasoning_content` and `thinking_blocks`.

### Steps to Reproduce

Send a POST request to the `/chat/completions` endpoint:

```json { "model": "gemma-4-31b-it", "messages": [ { "role": "user", "content": "what llm are you ??" } ], "think": false } ```

### Relevant log output

```shell The response incorrectly includes reasoning tokens and blocks:

{ "id": "0EzrabvyB4-fjMcPkpK8kQE", "model": "gemma-4-31b-it", "choices": [ { "finish_reason": "stop", "index": 0, "message": { "content": "I am a large language model, trained by Google.", "role": "assistant", "reasoning_content": "... (this should not be here) ...", "thinking_blocks": [ { "type": "thinking", "thinking": "..." …