Frames and Fans
Live
Components

AMD Project Zenith Preconfigures Windows 11 for AI Development

AMD has launched Project Zenith, a preconfigured Windows 11 developer environment for its Ryzen AI Halo platform, aiming to eliminate initial setup time for local AI work.

AMD has launched Project Zenith, a preconfigured Windows 11 developer environment for its Ryzen AI Halo platform, aiming...

AMD has made its Project Zenith developer environment available, a ready-to-code Windows 11 configuration preinstalled on the new Ryzen AI Halo Developer Platform. The system is powered by the Ryzen AI Max+ PRO 495 processor, and the software was developed in collaboration with Microsoft.

Project Zenith provides a preconfigured Windows 11 baseline with integrated development tools, allowing developers to start running AI workflows immediately without manual installation. This is a software and platform update rather than a new processor launch. The aim is to get an owner from unpacking the system to running a known AI workflow without having to assemble each software layer independently. AMD says the prepared environment is also coming to other Ryzen AI Max systems, but has not given a timetable for that wider rollout.

The platform supports large AI models with up to 300 billion parameters

The Ryzen AI Max+ PRO 495 processor enables local AI development with support for models up to 300 billion parameters, depending on format and quantization. The hardware foundation is the Ryzen AI Halo platform, which AMD specifies with up to 192 GB of unified memory, up to 160 GB of that pool available to graphics and AI tasks, and 273 GB/s of memory bandwidth. That is a capacity claim rather than a blanket statement that such models will run quickly. Context length, quantization, runtime overhead and inference speed all affect practical usability.

The processor combines 16 Zen 5 CPU cores, 32 threads, and boost frequencies reaching 5.2 GHz with integrated Radeon 8065S graphics and an XDNA 2 NPU. Its Radeon 8065S integrated graphics incorporates 40 RDNA 3.5 compute units, delivering up to 61 FP16 TFLOPS, alongside an XDNA 2 NPU rated for 55 TOPS.

Preconfigured tools eliminate initial setup time for developers

The development tools included are Visual Studio Code, Git, GitHub CLI, GitHub Copilot CLI, Azure CLI, Python, Node.js, .NET, PowerShell 7, and Windows Subsystem for Linux (WSL). The company says the setup provides a developer-ready Windows 11 baseline and integrates with its Ryzen AI Developer Center, which centralizes access to validated software versions, developer materials and guided workloads.

AMD AI Playbooks provide guided workflows for running large language models using PyTorch and ROCm, image generation through ComfyUI, local coding assistance using Qwen3-Coder, LLM deployment through LM Studio, and automation with n8n. AMD says Playbooks will be expanded over time. The supported workflows are useful demonstrations, but they do not guarantee that every model will fit in memory or that each tool will deliver the same performance as an NVIDIA CUDA system.

Full customization and framework flexibility are supported

AMD describes this as a starting point rather than a locked workflow. Developers can modify the environment and use any framework or language, avoiding vendor lock-in while maintaining AMD-validated software versions. Developers can customize the installation according to their requirements. For existing Ryzen AI Halo owners, AMD points to Microsoft’s Windows Developer Config project as a way to reproduce the software setup on a system that already has Windows installed.

The GitHub repository makes the setup script public, but users should review its actions and choose the configuration appropriate to their system before running it, since it adjusts Windows settings and installs development components. Project Zenith was originally announced by Microsoft during IFA 2026 in September. AMD says Ryzen AI Halo systems with Project Zenith pre-installed are available through Micro Center and its online store. No separate software fee was announced, and the blog does not state a new hardware price.

Related coverage

More from Components