Skip to content
Documentation menu

Share your model · Linux

Run a model on Linux.

Ompera downloads and runs a model on your computer. Use it in your apps, share it with your pools or make it available to everyone on Ompera.

Download DEB

Linux x86_64 · checking the latest package…

  • Ubuntu 22.04+, Debian 12+ or Linux Mint 21+, on an x86_64 computer
  • An Ompera account and internet access for the first download
  • Free disk space and memory for your model; a graphics card is optional

  1. Install the DEB

    Download the package above. Open a terminal in the folder where you saved it and run:

    sudo apt install ./ompera-provider-latest.deb

    APT installs the app and its required libraries. Then open Ompera Provider from your applications menu.

    Ompera opens and asks you to link this computer.
    Saved under another filename, or need to start from a terminal?

    Replace ompera-provider-latest.deb with the filename you downloaded, keeping ./ before it. To open the app from a terminal, run ompera-provider.

  2. Link your computer

    Open Workspace → Devices ↗ and click Connect a phone. That button also creates codes for the Linux app. Copy the code into Ompera on your computer and click Link this computer.

    The app shows the models you can run on this computer.
    The code expired?

    A code works once and expires after 10 minutes. Create a new one in Devices. This links the computer to your account; it is not an API key.

  3. Choose a model and who can use it

    Keep Run inside Ompera selected and choose a model that fits your computer. The suggested model takes its available memory into account. For a small first test, choose Qwen3.5 0.8B.

    Choose whether to share with your pools or everyone on Ompera. Your own use is free.

    You chooseA model
    It runs onThis computer
    Use it fromYour apps
    Already using LM Studio or Ollama?

    Choose Use my own server and enter the URL of the running OpenAI-compatible server. This is an Unshielded connection: the server owner can read requests. See Privacy & security.

  4. Start sharing

    Click Start sharing. The first start downloads the model and its engine, then checks that the model can answer requests. Wait for Sharing.

    Your model appears in Workspace → Models with an address you can copy.

    To use it from another app, create an API key and connect your tools. Keep Ompera open while sharing; no Chrome tab or browser extension is needed.

Update from the app

Click Check for updates at the bottom of the window. When an update is available, choose Update and restart. Ompera checks the signed package and asks for administrator approval to install it.

Updating stops sharing. Your models, settings and account link stay on the computer; start sharing again after the app reopens.

Updating an older app without this button?

Download the current DEB above, close Ompera and install it with the same APT command. Your saved models and settings are kept.

Graphics card missing, or the model will not start?

Open About this computer to check the detected graphics cards and driver. A selected driver package is not enough: its driver must be active. Use your distribution's driver manager; do not disable Secure Boot just to try the app.

In settings, select Processor only for a CPU check. After stopping sharing, About this computer → Test engine runs a small diagnostic model without sharing it or changing your selected model.

Use Send error report beside the error or in About this computer. You receive a report number; you do not need to copy logs manually. For other problems, send feedback.

Stop sharing or close the app

Click Stop sharing to stop providing requests. Closing the window also stops sharing and exits Ompera's model engine. Downloaded models stay on disk. A separate LM Studio or Ollama server keeps running independently.

This is a beta. Availability depends on your computer, drivers and network. Keep the computer awake while sharing. Running inside Ompera does not guarantee privacy from someone who controls the computer; see Privacy & security.