Quick Start

This section explains how to install LlamaPi and run your first model.

Install LlamaPi

If the Firefly APT repository is configured on the device, install LlamaPi directly:

sudo apt install llamapi

Alternatively, download the deb packages from the Firefly website, copy them to the device, open their directory, and install them locally:

sudo dpkg -i ./firefly-llamapi-*.deb

Use LlamaPi to Run a Model

Run the qwen3.5:4b model with the llamapi run command:

llamapi run qwen3.5:4b

The llamapi run command automatically selects and downloads a model available on the current hardware:

After the model is downloaded and loaded, the terminal enters an interactive chat. Chat with the model at the prompt:

For models with multimodal capabilities, use @ to attach files in the conversation:

Use /exit or Ctrl+D to leave the chat.

View Models Available on the Current Hardware

List models available on the current hardware:

llamapi list --online

The following example uses an RK3588 + RK1828 hardware platform:

Next Steps

• Download and remove local models: Download and Manage Models
• Run and deploy models: Run and Deploy Models
• Configure automatic loading at service startup: Persistent Deployment
• Connect a model to an existing application: Connect Third-Party Applications
• Use LlamaPi from a PC with a graphical interface: Using the Client