Getting started
Lantrn runs as two containers: the app and Qdrant for the vectors.
Install
Download the compose file into a new folder and start it:
curl -O https://raw.githubusercontent.com/Viedin/lantrn/main/Lantrn/docker-compose.yml
docker compose up -dOpen http://localhost:8080 and register. The first account becomes the admin. After that, new users need an invite.
Connect an embedding model
Search needs an embedding model. Fill in Settings → Embeddings:
| Field | Example |
|---|---|
| Base URL | http://host.docker.internal:1234/v1 |
| Model | text-embedding-qwen3-embedding-4b |
| API key | Only if your endpoint needs one |
The base URL must end with the version segment (/v1). Inside Docker, localhost is the container itself, so use host.docker.internal to reach LM Studio or Ollama on the host.
A vision model (for images) and a chat model (for answers) are optional. See Configuration.
Add documents and search
Create a collection on Collections, then upload files on Ingest or crawl a site on Crawl. See Adding documents.
Updating
docker compose pull
docker compose up -dYour data lives in Docker volumes and is kept. To stay on a version, pin the tag in docker-compose.yml, for example ghcr.io/viedin/lantrn:1.2.