MiniChat

A minimalistic LLM chat UI, inspired by Fooocus and Pi.

If you have a CORS enabled API endpoint to use, you can try it here.

The goal is to focus on the content and hide away as much tinkering and tweaking as possible.

We emphasize the content by using bold typography and minimal design.

Configuration

We try to keep configuration minimal yet flexible.

MiniChat supports OpenAI compatible APIs, see the Docker Compose section for running MiniChat with various model providers by leveraging gateways such as Portkey.

Docker

You can run MiniChat with Docker, the image is less than 2MB.

docker run -p 3216:3216 ghcr.io/functorism/minichat:master

Docker Compose

Ollama

Run

docker compose -f docker-compose-ollama.yml up
docker exec minichat-ollama-1 ollama pull llama3:8b

UI Settings:

• Model: llama3:8b
• OpenAI API URL: http://localhost:11434/v1
• API Key: unused-for-ollama

Portkey

Run

docker compose -f docker-compose-portkey.yml up

UI Settings:

• Model: claude-3-5-sonnet-20240620
• OpenAI API URL: http://localhost:8080/http://portkey:8787/v1
• API Key: sk-...
• Headers
  • x-portkey-provider: anthropic

HuggingFace Text Generation Inference

Run

docker compose -f docker-compose-tf-tgi.yml up

UI Settings:

• Model: HuggingFaceH4/zephyr-7b-beta
• OpenAI API URL: http://localhost:8080/v1
• API Key: unused-for-hf-tgi
• Headers
  • Check "Send Minimal Headers"