LLM Aggregator

This is a very simple personal use LLM aggregator that helps in passing same prompts to multiple LLMs and then viewing their output in UI and then take a call which one is better.

Supported LLMs:

1. ChatGPT
2. Gemini
3. Claude
4. Llama

How to run

1. Setup Postgres:

1. Install docker
2. Start potgres container using below command from infra directory:

   docker compose up --build -d
   

   • On first run this command will automatically create schemas
   • Postgres data will be stored in pg_data folder so data will not be lost even if container stops.

2. Start Backend

1. Use node 21

2. Get key for ChatGPT, Gemini and Claude and set them in the environment. (.env file in repository root can also be used)

   GEMINI_API_KEY=<your_key>
   CHATGPT_API_KEY=<your_key>
   CLAUDE_API_KEY=<your_key>
   

3. Run llama locally (ollama can be used to run llama locally)

4. npm install

5. npm run execute

3. Start Frontend

For Development:

1. npm install
2. npm start

For Usage:

1. npm i -g serve - needs to be done only first time
2. npm run build
3. serve -s build

Demo

demo.mp4