LLM Stack

An all-in-one Docker Compose config for providing access to local and external LLMs with multiple chat interfaces.

Components

• Caddy: Acts as central entrypoint for the whole stack
• Ollama: Provides access to local LLM models
• LiteLLM: OpenAI compatible API proxy for local Ollama provided models and upstream models
• Multiple ChatGPT-style web interfaces for interacting with the LLM models

Models

• Local
  • local-mistral
  • local-mixtral-8x7b
  • local-llama3-8b
• OpenAI
  • openai-gpt-3.5-turbo
  • openai-gpt-4-turbo
  • openai-gpt-4o
• Google
  • google-gemini-1.5-pro
• Anthropic
  • anthropic-claude-3-sonnet
  • anthropic-claude-3-opus
• Groq
  • groq-llama3-70b

Chat Frontends

• Open WebUI
• NextChat
• YakGPT

Getting Started

Prerequisites

• Docker
• Docker Compose
• Git

Setup

1. Clone this repository
2. Copy the default config cp default.env .env
3. Edit .env and add the relevant API keys
4. Start the Docker Compose configuration: docker-compose up
5. Access the Caddy webserver at http://localhost:3000