Support for Streaming Responses with Tool Calling in langchain_ollama

[Kana2581]

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

Support for streaming responses with tool calling in langchain_ollama.

Ollama v0.8.0 recently added native support for tool calling along with streaming responses:
→ https://github.com/ollama/ollama/releases/tag/v0.8.0
However, the current langchain_ollama integration does not yet support this functionality.

It would be great if langchain_ollama could support tool calling and streaming output at the same time, enabling more responsive and interactive agent behaviors in LangChain.

Motivation

I'm building a tool-using agent that relies on streaming output for responsiveness. While Ollama now supports tool calling with streaming, I can't take advantage of this in LangChain due to the current limitations in langchain_ollama.

Streaming output dramatically improves UX in chat UIs, and combining this with tool use (e.g., function-calling or retrieval plugins) is becoming essential.

This issue might be related to or blocked by the current implementation of the Ollama integration — if there's already ongoing work or a plan for this, I'd appreciate any update or guidance. Happy to help with testing or a potential PR!

Proposal (If applicable)

No response

[mdrxy]

Implemented! Make sure you're on the latest Ollama binary.