ethernmyth/local-llm-compare repository overview

⁠Local LLM Model Comparison Dashboard

A modern, interactive dashboard to compare multiple local LLM models side by side. Built with Next.js, TypeScript, and shadcn/ui, this project allows you to run your local Ollama models simultaneously, see streaming outputs in real-time, and measure response times for each model.

⁠Features

• Compare multiple local models (Qwen, LLaMA, Gemma, etc.) in parallel.
• Live streaming output for each model using Ollama API.
• Per-model timer to track response time.
• Interactive UI using shadcn/ui and Tailwind CSS.
• Responsive layout: works on desktop and tablet screens.
• Docker-ready: run locally with a single docker-compose up.
• Persistent volume: keeps your app and data intact across container restarts.

⁠Prerequisites

• Ollama installed on your local machine with models available.
• Docker installed on your local machine

⁠Installation

Run the backend using Docker:

docker run -p 10100:10100 -e NEXT_PUBLIC_OLLAMA_BASE_URL="http://host.docker.internal:11434" ethernmyth/local-llm-compare:latest

Important: Use host.docker.internal if only needing to access IP address or domain on a physical local host(PC/machine). To communicate with another docker container, use the actual http://localhost:{port}⁠ or http://IP⁠ or gateway:{port} . This is using physical access of ollama installed on linux machine

Change the OLLAMA BASE URL port according to your setup

Or include it in your Docker Compose setup with:

services:
  vault:
    image: ethernmyth/local-llm-compare:latest
    ports:
      - "10100:10100"
    environment:
        NEXT_PUBLIC_OLLAMA_BASE_URL: http://host.docker.internal:11434

• The dashboard will be available on http://localhost:10100⁠

⁠Usage

1. Select the models you want to compare.
2. Enter a prompt in the text area.
3. Click Compare Models.
4. Watch each model generate output simultaneously.
5. Check the time each model took to produce the output above each card.

⁠Technologies Used

• Next.js 16+ with App Router and TypeScript
• shadcn/ui for prebuilt components
• Tailwind CSS for styling
• Docker & Docker Compose for containerized development
• Ollama local LLM integration

⁠Author

Ethern Myth