AMD Introduces Optional Adrenalin AI Bundle to Simplify Local AI Development on Windows
The new AMD Software: Adrenalin Edition 26.1.1 drivers introduce a major update called AI Bundle, an optional feature announced during CES 2026. This new package is designed to simplify and speed up local AI setup, removing cloud dependency and taking advantage of on-device acceleration through technologies such as HIP SDK, Vulkan, and ROCm.
With a single simplified installation, AI Bundle equips AMD-based systems with essential tools to start creating and running AI workloads. It eliminates complex setup steps and reduces installation time. More importantly, it transforms Adrenalin from a hardware monitoring and tuning tool into a complete AI platform.
Users can now access popular applications for image generation and local large language models (LLMs), as well as new PyTorch support for Windows, making it easier than ever to experiment with AI development directly on their PC.
A Central Hub for Local AI Applications
AI Bundle acts as a central hub for AI apps, accessible through a dedicated “AI” tab under “Apps” in the Adrenalin interface. The optional installer, roughly 34 GB in size, includes third-party tools and requires users to accept the related license agreements (EULA).
The AI Bundle option appears during the Adrenalin Edition installation. It supports optional telemetry and includes Python 3.12 as the default interpreter. Once installed, users can launch AI applications directly from the AMD software with simple workflows — from running Ollama prompts to creating PyTorch virtual environments with ROCm.
The AI Bundle includes the following tools:
- PyTorch on Windows – Build and train models from scratch or load exist models for inference
- ComfyUI – Visual workflows for image generation, accelerated by ROCm PyTorch
- Ollama – Run LLMs locally for text generation and prototyping
- LM Studio – Install and run large language models with easy parameter tuning
- Amuse – AMD optimized text-to-image generation
To run these tools locally, users need an AMD Radeon RX 9000 or RX 7000 GPU (RX 7700 or higher) or systems equipped with Ryzen AI Max, Ryzen AI 400, or Ryzen AI 300 processors with NPUs.
