Skip to main content

Access open source LLMs in your local machine with CLI support

Project description

delta logo

Delta

nan corp inference file logo

Access open source LLMs in your local machine

Announcement Blog Post

Introduction to Delta

In a world where cutting-edge AI often demands expensive hardware like M4 chips or DGX systems, Delta stands out as a game-changer, particularly for regions like Africa where many rely on older laptops with CPUs. This project is personal—it's about bridging the gap for those left behind by the rapid pace of technological advancement, empowering users to harness the power of large language models (LLMs) without needing costly devices. Delta transforms local AI inferencing with a single-file engine that runs open-source LLMs on your CPU using just a terminal command. With CLI support, it delivers efficiency, fast backend runtime with zero overhead and optimized memory usage, making powerful AI accessible on standard hardware. What sets Delta apart is its ability to support distributed inferencing—not from the cloud, but across local computers—unlocking their collective potential seamlessly. With Delta, you can chat with LLMs offline, search Wikipedia, GitHub, arXiv, the internet via DuckDuckGo, or your own files, and even control your system with natural language commands like creating files or opening apps. Designed to go beyond tools like vLLM, llama.cpp, llm.c, Jan, LMStudio, and Ollama, Delta brings AI to your fingertips with simplicity and flexibility, aiming to democratize access and close the global AI divide.

macOS Windows Linux

Delta Installation Guide

Delta is an AI assistant that runs locally on your machine. This guide will help you install Delta on your preferred operating system.

Prerequisites

Before installing Delta, ensure your system meets these requirements:

  • 4GB RAM minimum (8GB or more recommended)
  • 10GB free disk space
  • Internet connection for initial setup

Installation Instructions

Quick Install with pip (Recommended)

The easiest way to install Delta is using pip:

pip install delta-ai

After installation, you can use Delta immediately:

# Check if installation was successful
delta --help

# List available models
delta list

# Run a model (will download if not available)
delta run mistral

Note: This method requires Python 3.8+ and pip to be installed on your system.

Alternative Installation Methods

If you prefer not to use pip or need a custom installation, you can use the platform-specific installers below.

Windows

Coming Soon ...

macOS

  1. Download delta δ Installer

i. Using Link:

  • Download δ
    • This will download delta-installer-v1-macos.sh

ii. Open Terminal and run:

  • In Terminal navigate to the folder with delta-installer-v1-macos.sh
    cd navigate/to/the/folder/with/delta-installer-v1-macos.sh
    

then

  • Run the installation script
    chmod +x delta-installer-v1-macos.sh
    ./delta-installer-v1-macos.sh
    
  1. Dowload and Install via Terminal Open terminal
  • Make sure you have curl

    curl --version
    
  • If you see curl version continue to the next step if not, download curl.

    brew install curl
    
  • Download the installer

    curl -O https://nileagi.com/delta-installer-v1-macos.sh
    
  • Make the script executable

    chmod +x delta-installer-v1-macos.sh
    
  • Run the installation script

    ./delta-installer-v1-macos.sh
    

Linux

  1. Download delta δ Installer

i. Using Link:

  • Download δ
    • This will download delta-installer-v1-linux.sh

ii. Open Terminal and run:

  • In Terminal navigate to the folder with delta-installer-v1-linux.sh

    cd navigate/to/the/folder/with/delta-installer-v1-linux.sh
    

then

  • Run the installation script
    chmod +x delta-installer-v1-linux.sh
    ./delta-installer-v1-linux.sh
    
  1. Dowload and Install via Terminal Open terminal
  • Make sure you have curl

    curl --version
    
  • If you see curl version continue to the next step if not, download curl.

    sudo apt update && sudo apt install curl -y
    
  • Download the installer

    curl -O https://nileagi.com/delta-installer-v1-linux.sh
    
  • Make the script executable

    chmod +x delta-installer-v1-linux.sh
    
  • Run the installation script

    ./delta-installer-v1-linux.sh
    

The installer will:

  • Install Homebrew if not present - for macos
  • Install Python 3 if not present
  • Install Git if not present
  • Install Ollama if not present
  • Set up Delta in your home directory

After installation:

  • Open a new Terminal window
  • Type delta to start the application
  • The first run will download the default AI model (deepseek-r1:1.5b)

Quickstart


To run and chat with deepseek-r1:1.5b:

delta run deepseek-r1:1.5b

Troubleshooting

Common Issues

  1. "Command not found: delta"

    • Try restarting your terminal
    • On Windows, restart your computer to ensure PATH changes take effect
  2. Installation fails

    • Ensure you have administrator privileges
    • Check your internet connection
    • Make sure you have enough disk space
  3. Ollama not starting

    • On Windows: Run ollama serve in a separate terminal
    • On macOS/Linux: Run sudo systemctl start ollama

Getting Help

If you encounter any issues:

  1. Check the FAQ
  2. Visit our GitHub repository
  3. Contact support at support@nileagi.com

Model library


delta support a list models available on ollama.

Here are some examples of the models that can be downloaded

Model Parameter Size Dowload
Gemma 3 1B 815MB delta pull gemma3:1b
Gemma 3 4B 3.3GB delta pull gemma3
Gemma 3 12B 8.1GB delta pull gemma3:12b
DeepSeep-R1 7B 4.7GB delta pull deepseek-r1
Llama 3.2 3B 2.0GB delta pull llama3.2
Llama 3.2 1B 1.3GB delta pull llama3.2:1b
Llama 3.1 8B 4.7GB delta pull llama3.1
Phi 4 14B 9.1MB delta pull phi4
Phi 4 Mini 3.8B 2.5GB delta pull phi4-mini
Mistral 7B 4.1GB delta pull mistral
Moondream 2 1.4B 829MB delta pull moondream
Neural Chat 7B 4.1GB delta pull meural-chat
Starling 7B 4.1GB delta pull starling-lm
Code Llama 7B 3.8GB delta pull codellama
Llama 2 Uncensored 7B 3.8GB delta pull llama2-uncensored
LLaVA 7B 4.5GB delta pull llava
Granite-3.3 8B 4.9GB delta pull granite3.3
You should have at least 8GB of RAM available to run the 7B models, 16 GB to run 13B models, and 32 GB to run the 33B models.

Use Delta


Help

delta -h

Install model

Syntax delta pull model_name

delta pull llama3.2:1b

Check Models installed

delta list

Remove / Delete model

Syntax delta remove model_name

delta remove llama3.2:1b

Chat with the Models only

Syntax delta run model_name

delta run llama3.1

Search through Wikipedia then use model to answer

Syntax delta run model_name --wiki

delta run llama3.1 --wiki

search through DuckduckGo (Browser) then use model to answer

Syntax delta run model_name --ddg

delta run llama3.1 --ddg

Search through your local documents or summarize

Syntax delta run model_name --docs /location/to/your/document

Linux or MacOs

delta run llama3.1:latest --docs /home/Documents/computer_vision_nlp.pdf

Windows

delta run llama3.1:latest --docs C:\Users\home\Documents\computer_vision_nlp.pdf

Check your sys specs and Model Recomendations

delta check

View your Chat History

delta hist

Clear Chat History

delta hist --clear

Delta Version

delta --version

Run Delta

At the > prompt

  • Enter: Adds a new line.

  • Shift + Arrow Keys: Select text (left, right, up, down).

  • Ctrl + C: Copies selected text to the clipboard or exits if no selection.

  • Ctrl + V: Pastes clipboard content at the cursor.

  • Ctrl + j: Submits the input.

  • Ctrl + D: or type "exit"Exits the session.

  • Arrow Keys: move your coursor left, right, up and down on delta.

Uninstallation

Windows

  1. Delete the .delta folder in your user directory
  2. Remove Delta from your PATH environment variable

macOS/Linux

  1. Delete the .delta folder in your home directory
  2. Remove the Delta entry from your shell's configuration file (.bashrc, .zshrc, etc.)

System Requirements

  • Operating System:

    • Windows 10/11
    • macOS 10.15 or later
    • Ubuntu 20.04 or later, or equivalent Linux distribution
  • Hardware:

    • CPU: 2GHz dual-core processor or better
    • RAM: 4GB minimum (8GB recommended)
    • Storage: 10GB free space
    • Internet: Required for initial setup and model downloads

Privacy and Security

Delta runs completely locally on your machine. No data is sent to external servers except for:

  • Initial package downloads
  • Model downloads from Ollama
  • Optional telemetry (can be disabled)

License

Delta is licensed under the MIT License. See the LICENSE file for details.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

delta_ai-1.0.0.tar.gz (128.3 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

delta_ai-1.0.0-py3-none-any.whl (105.9 kB view details)

Uploaded Python 3

File details

Details for the file delta_ai-1.0.0.tar.gz.

File metadata

  • Download URL: delta_ai-1.0.0.tar.gz
  • Upload date:
  • Size: 128.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/2.1.3 CPython/3.11.13 Darwin/24.3.0

File hashes

Hashes for delta_ai-1.0.0.tar.gz
Algorithm Hash digest
SHA256 d2604411b52160abf1f3ad2913e64c3fb5bed3f5a626216ceebbadad538fc273
MD5 4fc3e5d51de4274b392b97f347b28559
BLAKE2b-256 e2a713cb5989c6d4a63d119ed089bcf122516c95912070c00fd8b42cc46978e3

See more details on using hashes here.

File details

Details for the file delta_ai-1.0.0-py3-none-any.whl.

File metadata

  • Download URL: delta_ai-1.0.0-py3-none-any.whl
  • Upload date:
  • Size: 105.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/2.1.3 CPython/3.11.13 Darwin/24.3.0

File hashes

Hashes for delta_ai-1.0.0-py3-none-any.whl
Algorithm Hash digest
SHA256 e7c8127109bb40a7e388111b4f5f9020e19a87fc3146bc4783c26e2b51a7d6ad
MD5 5e57ade4993daaa129e76b65c00a7332
BLAKE2b-256 3c8520d297967857b2b5f1e34aa4f42cc4bb715e868c73366987ce3473b02742

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page