Dell Pro AI Studio (`dpais`) CLI

Overview

The Dell Pro AI Studio (dpais) CLI is a powerful command-line tool designed to manage and run AI models locally on your Dell hardware. It automates the setup of necessary services, intelligently filters models based on your system’s hardware (CPU, GPU, NPU), and provides a simple interface for installing, running, and managing a catalog of curated AI models.

Pre-release Software Please note this is a dev build and is pre-release.

Key Features:

Automated Setup: A single init command installs core services, dependencies, and AI models.
Hardware-Aware: Automatically detects your system’s architecture (x64/ARM64) and compute devices (CPU, NPU, GPU) to show and install compatible models.
Diverse Model Catalog: Supports various model types including Large Language Models (LLMs), Transcription, Image Embedding, and Image Upscaling.
Interactive Chat: Run and interact with LLMs directly in your terminal.
Performance Benchmarking: Test the performance of installed models on your hardware.
Service Management: Easily check the status of and restart the underlying AI runtime service.

Getting Started: Installation and Initial Setup

The init command is the starting point. It prepares your system by installing all required dependencies, core services, and then guides you through installing your first AI models.

Prerequisites:

Windows Operating System
Administrator privileges (the CLI will prompt for elevation if needed)
An active internet connection

Install here:


winget install Dell.DPAIS

Download

Run the `init` command

Open your terminal and run:


dpais init

This command will:

Check for Administrator Privileges: If not running as an administrator, the CLI will prompt you to relaunch with elevated permissions.
Install Dependencies: It uses winget to automatically install required runtimes like .NET 8 and the VC++ Redistributable. You can skip this with the --skip-deps flag.

Important Dependency Note Please note that your .NET Core and Desktop runtimes must be the same version. The CLI handles this, but if you are installing dependencies manually, be aware of this requirement.
Install Core Services: Downloads and installs the core “Dell AI Framework” and “DellTechHub” service, which manage the AI models in the background.
Install Models: After the core services are installed, it will present a list of compatible models for you to install.

Installation for Core Services Only

If you only want to install the background services without being prompted to install models immediately, use the --core flag:


dpais init --core

You can then install models later using the dpais model install command.

Core Commands

These commands help you manage the DPAIS environment and view system information.

`compute` (or `hardware`)

Displays a list of AI-capable compute devices (CPU, GPU, NPU) detected on your system. This is useful for understanding which models will be available for you to install.


dpais compute

`ping`

Checks the status of the DPAIS runtime service (DellTechHub). This is the quickest way to see if the AI backend is ready.


dpais ping

`restart`

Restarts the DPAIS runtime service. This is useful if the service becomes unresponsive or after installing new models to ensure they are loaded correctly.


dpais restart

Managing Models (`dpais model`)

The model command is the primary way you will interact with the AI models themselves.

`model list` (or `ls`)

Lists the AI models available for your system, indicating which are already installed.

To see a list of compatible models for your system:
```
dpais model list
```
To see a more detailed view:
```
dpais model list --verbose
```
To see the entire model catalog:
```
dpais model list all
```

`model install`

Downloads and installs one or more AI models.

Launch an interactive selection prompt:
```
dpais model install
```

Install a specific model directly by name:


dpais model install --model "Phi-3.5-mini-instruct"

`model run` (or `chat`)

Starts an interactive chat session with an installed Large Language Model (LLM).

To choose from installed LLMs:
```
dpais model run
```
To run a specific LLM:
```
dpais model run --model phi3.5
```
To start a chat with a custom system prompt:
```
dpais model run --model phi3.5 --system "You are a pirate."
```
Type exit to end the chat session.

`model performance`

Runs a benchmark test on an installed model to measure its performance on your hardware.


dpais model performance --model whisper

If the model option is not provided, it will prompt you to choose one. The output includes metrics like load time, average execution time, and (for LLMs) tokens per second.

`model uninstall`

Uninstalls one or more AI models from your system.

To launch an interactive selection prompt:
```
dpais model uninstall
```
To uninstall a specific model:
```
dpais model uninstall --model phi3.5
```
To uninstall all currently installed models:
```
dpais model uninstall --model -all
```

Complete Uninstallation

The uninstall command removes the Dell Pro AI Studio Framework and, optionally, all models and related Dell services.

To uninstall all models and the Dell AI Framework:
```
dpais uninstall
```
To also uninstall the Dell Core Services (DellTechHub):
```
dpais uninstall --dcs
```
To leave models untouched and only remove the Dell AI Framework:
```
dpais uninstall --skip-model-uninstall
```

Dell Pro AI Studio (dpais) CLI

Overview

Getting Started: Installation and Initial Setup

Install here:

Run the init command

Installation for Core Services Only

Core Commands

compute (or hardware)

ping

restart

Managing Models (dpais model)

model list (or ls)

model install

model run (or chat)

model performance

model uninstall