Open Model Chat

Open Model Chat Backend is a TypeScript + Express service that wraps local Ollama models and exposes JSON + streaming endpoints for chat and vision use cases. A frontend application Open Model Frontend was developed to utilise this application's capabilities.

Features

• Chat with a local Ollama model.
• Stream chat responses via Server-Sent Events (SSE).
• Vision endpoint that accepts an uploaded image and streams the response.
• List available local Ollama models with additional details.

Tech Stack

• Node.js + TypeScript
• Express + CORS
• Multer (file uploads)
• Ollama JS client

Prerequisites

• Node.js 18+ (or newer)
• npm
• Ollama installed and running locally

Local Setup

1. Install dependencies:

   npm install

2. Start the dev server:

   npm run dev

3. Build and run in production mode:

   npm run build
   npm start

The server listens on PORT (default 3000).

Ollama Setup (Local)

1. Install Ollama for your OS:

   • macOS: install the Ollama app.
   • Linux: use the official install script.
   • Windows: use the Ollama installer.

2. Start the Ollama service.

3. Pull the default model used by this backend:

   ollama pull gemma3:4b

4. Verify Ollama is running:

   ollama list

If you want to use a different model, pass model in the request body to the chat/stream/vision endpoints.

Model Switching

You can install any Ollama model you want locally and switch models per request by sending the model field in the payload. The backend simply forwards the model name to Ollama, so whatever you have pulled locally can be used.

Install a model:

ollama pull llama3.1:8b

Use that model in a request:

{
  "model": "llama3.1:8b",
  "messages": [
    { "role": "user", "content": "Explain the architecture." }
  ]
}

Notes:

• If model is omitted, the server defaults to gemma3:4b.
• Use ollama list to see which models are available locally.

API Overview

Base path: /api/ai

Response Format

All JSON responses use this shape:

{
  "status": true,
  "message": "Response from AI model",
  "data": "..."
}

On error:

{
  "status": false,
  "message": "Error message",
  "error": "..."
}

POST /api/ai/chat

Send a single prompt and receive a full response.

Request body:

{
  "message": "Say hello"
}

Example:

curl -X POST http://localhost:3000/api/ai/chat \
  -H "Content-Type: application/json" \
  -d '{"message":"Say hello"}'

POST /api/ai/stream

Stream tokens using SSE. You can specify a model and pass a full messages array.

Request body:

{
  "model": "gemma3:4b",
  "messages": [
    { "role": "user", "content": "Summarize this project." }
  ]
}

Example:

curl -N -X POST http://localhost:3000/api/ai/stream \
  -H "Content-Type: application/json" \
  -d '{"messages":[{"role":"user","content":"Summarize this project."}]}'

The server streams SSE events as:

data: {"token":"..."}

data: [DONE]

POST /api/ai/vision

Upload an image and stream a response (SSE).

Form fields:

• image (file, required)
• prompt (string, optional)
• model (string, optional)

Example:

curl -N -X POST http://localhost:3000/api/ai/vision \
  -F "image=@/path/to/image.jpg" \
  -F "prompt=Describe the image."

GET /api/ai/getAvailableModelsWithDetails

Returns local Ollama models along with capability info.

Example:

curl http://localhost:3000/api/ai/getAvailableModelsWithDetails

Notes

• Uploaded files are stored in uploads/.