Run a model in the background, and stop it

By default, tensorless run stays in the foreground: it shows the server’s logs, and Ctrl-C stops the server. With --detach it returns as soon as the model has loaded and leaves the server running, so you can close the terminal.

You need the launcher installed and a host that passes tensorless doctor. See Prepare a Linux host for a P100.

1. Start the server in the background

tensorless run iamMrDuncan/Qwen3.8-27B-PN3-GGUF --detach

run downloads and checks whatever is missing first, as it always does. Then it starts the server’s container, waits for the model to load, and returns. Loading takes a few seconds once the file is in the page cache, and longer the first time. While it waits it prints:

waiting for http://127.0.0.1:8080/health while the model loads (up to 20m)

When the server answers, it prints the endpoint and how to stop it:

iamMrDuncan/Qwen3.8-27B-PN3-GGUF is ready at http://127.0.0.1:8080 (OpenAI-compatible API under /v1)
try it:
  curl http://127.0.0.1:8080/v1/chat/completions -H 'Content-Type: application/json' \
    -d '{"model": "iamMrDuncan/Qwen3.8-27B-PN3-GGUF", "messages": [{"role": "user", "content": "Hello"}]}'
stop it with: tensorless stop iamMrDuncan/Qwen3.8-27B-PN3-GGUF

If the image and the model are already pulled, tensorless serve iamMrDuncan/Qwen3.8-27B-PN3-GGUF --detach starts the server without pulling or downloading anything. It still checks both, and fails if either is missing.

Note

If the server stops before it is ready, --detach prints the last 60 lines of its log and fails. Fix a failed check or a failed start says what the usual lines mean.

2. Read the logs

The server runs in a container named tensorless- followed by the model’s repo id in lower case, with / replaced by -:

Model	Container
iamMrDuncan/Qwen3.8-27B-PN3-GGUF	tensorless-iammrduncan-qwen3.8-27b-pn3-gguf
iamMrDuncan/Qwen3.8-27B-UD-IQ3_XXS-Q3Floor-GGUF	tensorless-iammrduncan-qwen3.8-27b-ud-iq3_xxs-q3floor-gguf

Follow its log with the container tool that started it, podman or docker:

podman logs -f tensorless-iammrduncan-qwen3.8-27b-pn3-gguf
# or:
docker logs -f tensorless-iammrduncan-qwen3.8-27b-pn3-gguf

Ctrl-C stops following the log. The server keeps running.

3. Stop the server

tensorless stop iamMrDuncan/Qwen3.8-27B-PN3-GGUF

stopped tensorless-iammrduncan-qwen3.8-27b-pn3-gguf

stop looks for the container in both podman and docker, whichever are installed, so you don’t need to remember which one started it. The container was started with --rm, so stopping it also removes it. The model file and the image stay where they are, and the next run starts again in seconds.

Check that it worked

While the server runs, it answers on its health address:

curl -s http://127.0.0.1:8080/health

After tensorless stop, the same command fails to connect, and tensorless stop again says:

tensorless-iammrduncan-qwen3.8-27b-pn3-gguf is not running

The server does not start again by itself after a reboot. Run tensorless serve MODEL --detach again, or start it from your own service manager.