Adding a parameter to OllamaLLM to disable the thinking output.

[dhanesh24g]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

The Ollama LLMs are generating blocks by default. There's no straight-forward way to suppress this output. So kindly add a parameter in OllamaLLM to disable the thinking output.

Something like -

llm_qwen = OllamaLLM(
    base_url="http://localhost:11434",
    model="qwen3:latest",
    temperature=0.5,
    max_tokens=400,
   
    # Addition of think_mode
    think_mode=False
)

Motivation

To make it easier for the langchain_ollama users to work with LLM models.

Proposal (If applicable)

Something like -

llm_qwen = OllamaLLM(
    base_url="http://localhost:11434",
    model="qwen3:latest",
    temperature=0.5,
    max_tokens=400,
   
    # Addition of think_mode
    think_mode=False
)

[k-jayanth]

yes it would be great to see this feature

[mdrxy]

This is now supported by setting reasoning=False either in the class param or during invocation.

[k-jayanth]

Thanks @mdrxy

[dhanesh24g]

Thanks @mdrxy !