Ollama-API-Proxy

A proxy for Ollama to easily disable thinking output for Home Assistant integrations.

Features

• Proxies /api/chat requests to Ollama AI server, automatically setting think=false to disable thinking output.
• Streams responses back to the client.
• Supports fetching tags via /api/tags.
• Simple health check endpoint /.

Requirements

• Python 3.9+
• FastAPI, httpx, uvicorn

Setup

1. Install dependencies:

   pip install fastapi httpx uvicorn 
   
   

2. Configure the Ollama host URL in your code (OLLAMA_HOST).

3. Run the proxy server:

   uvicorn proxy:app --host 0.0.0.0 --port 11435
   
   

Usage

• POST /api/chat: Forward chat requests to Ollama with thinking output disabled.
• GET /api/tags: Retrieve available tags.
• GET /: Health check.