What is the Razer AI Kit?
The open-source SDK powering Project AVA and how AVA SDK extends its capabilities for developers and enthusiasts.
The Technological Heart of Project AVA
The Razer AI Kit (aikit) is the official open-source development kit created by Razer for Project AVA. Published under the MIT license on GitHub (razerofficial/aikit), this Python SDK lets developers and enthusiasts build fully local AI assistants that run directly on your PC without relying on cloud services.
The Razer AI Kit source code is available on GitHub: razerofficial/aikit
What is the Razer AI Kit?
The AI Kit is a Python-based SDK designed to run local language models with hardware acceleration. It provides a modular architecture with an intents system, plugins, and personality configuration — all running locally.
Local Inference
Run language models like Llama 3 directly on your NVIDIA GPU without sending data to the cloud.
NPU Acceleration
Leverage neural processing units to optimize AI performance in real-time.
Intents System
Intent-based architecture that allows the assistant to understand and execute complex commands.
Open Source
Fully open-source with MIT license. Inspect, modify, and contribute to the code freely.
Relationship with AVA SDK
AVA SDK is the complete ecosystem that wraps around the Razer AI Kit. While the AI Kit provides the technological core — the inference engine, plugin system, and intents framework — AVA SDK adds the layers needed for developers and creators to make the most of it: documentation, tutorials, customization tools, and community.
AVA SDK and the Razer AI Kit together form a complete ecosystem. The AI Kit is the engine; AVA SDK is the manual, toolbox, and community that helps you make the most of it.
Key Components
These are the three technological pillars of the Razer AI Kit. Learn about them in detail in our Core Architecture guide.
vLLM Engine
High-performance inference engine for language models. Optimized for local execution with dynamic memory paging and continuous batching.
Ray Framework
Distributed computing framework that orchestrates model execution and enables horizontal scaling of processing.
LlamaFactory
Fine-tuning tool that enables efficient LLM customization with your own data for specific use cases.
System Requirements
For a detailed step-by-step installation and setup guide, visit our Lesson 01: Execution Environment.
Getting Started
Clone the repository: git clone https://github.com/razerofficial/aikit
Set up the environment with Docker: docker compose up
Download a base model (Llama 3, Mistral, etc.)
Run your first inference with the included scripts
Explore AVA SDK documentation to customize AVA
Ready to get started?
Follow our step-by-step guide to prepare your development environment.
Start with Lesson 01Frequently Asked Questions
Is the Razer AI Kit free?
Yes, it is completely free and open-source under the MIT license. You can download, use, and modify it without restrictions from the official Razer repository on GitHub.
What hardware do I need to run the Razer AI Kit?
You need a PC with an NVIDIA GPU compatible with CUDA 12.1+, 16 GB of RAM, and at least 50 GB of free disk space. You do not need a specific Razer device — it runs on standard hardware.
Is the Razer AI Kit the same as Project AVA?
Not exactly. Project AVA is Razer's holographic product — an AI assistant with holographic projection. The Razer AI Kit is the open-source SDK that Project AVA is built on. You can use the AI Kit to create your own assistants without the holographic hardware.
Can I use the Razer AI Kit without a Razer device?
Yes, absolutely. The AI Kit works on any PC with a compatible NVIDIA GPU. It does not require any proprietary Razer hardware, making it accessible to any developer interested in local AI.