Skip to content

[Bug]: The content output of the gemini 3 model is sent to the thinking part. #16816

@LFSCODE

Description

@LFSCODE

What happened?

Input:
{
"model": "gemini-3-pro-preview",
"top_p": 0.9,
"stream": false,
"messages": [
{
"role": "user",
"content": "hello, introduce yourself"
}
],
"max_tokens": 5000,
"temperature": 0.9,
"reasoning_effort": "high"
}

Output:
{
"id": "T6QdadH-BaLnlOoP0vew8Qs",
"created": 1763550286,
"model": "gemini-3-pro-preview",
"object": "chat.completion",
"choices": [
{
"finish_reason": "stop",
"index": 0,
"message": {
"content": "Hello! I am Gemini, a large language model built by Google.\n\nI am designed to be a helpful and capable AI assistant. I can help you with a wide range of tasks, such as:\n\n* Answering questions: I can provide information on almost any topic.\n* Writing and creating: I can help draft emails, essays, stories, poems, and code.\n* Summarizing: I can condense long articles or documents into quick summaries.\n* Translating: I can translate text between many different languages.\n* Brainstorming: I can help generate ideas for projects or solve problems.\n\nHow can I help you today?",
"role": "assistant",
"images": [],
"thinking_blocks": [
{
"type": "thinking",
"thinking": "{"text": "Hello! I am Gemini, a large language model built by Google.\n\nI am designed to be a helpful and capable AI assistant. I can help you with a wide range of tasks, such as:\n\n* Answering questions: I can provide information on almost any topic.\n* Writing and creating: I can help draft emails, essays, stories, poems, and code.\n* Summarizing: I can condense long articles or documents into quick summaries.\n* Translating: I can translate text between many different languages.\n* Brainstorming: I can help generate ideas for projects or solve problems.\n\nHow can I help you today?"}",
"signature": "xxxxxx"
}
]
}
}
],
"usage": {
"completion_tokens": 957,
"prompt_tokens": 4,
"total_tokens": 961,
"completion_tokens_details": {
"reasoning_tokens": 815,
"text_tokens": 142
},
"prompt_tokens_details": {
"text_tokens": 4
}
},
"vertex_ai_grounding_metadata": [],
"vertex_ai_url_context_metadata": [],
"vertex_ai_safety_results": [],
"vertex_ai_citation_metadata": []
}

Relevant log output

Are you a ML Ops Team?

No

What LiteLLM version are you on ?

v1.75.5.rc.1

Twitter / LinkedIn details

No response

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions