AMD Ryzen AI Halo Developer Platform (Ryzen AI Max+ 395)

The AMD Ryzen AI Halo Developer Platform is AMD's own reference-hardware dev kit built around the Ryzen AI Max+ 395 Strix Halo processor. It runs local LLM inference and fine-tuning on x86 hardware with full ROCm support. It suits ML engineers and researchers whose toolchain, CI pipeline or model zoo assumes x86_64 and who want an AMD-native alternative to the NVIDIA GB10 systems.

Who it's for

• Running open-weight models in the 30B–70B parameter range locally
• Fine-tuning and inference on frameworks with solid ROCm support, such as PyTorch, llama.cpp and vLLM's ROCm backend
• Teams that depend on x86 Docker images and Python wheels
• Developers evaluating ROCm with ready-made setups for LM Studio, ComfyUI and VS Code

Key features

• AMD Ryzen AI Max+ 395 processor: 16 Zen 5 CPU cores / 32 threads at up to 5.1GHz boost (3.0GHz base), with 64MB of L3 cache, built on a 4nm process.
• Integrated Radeon 8060S GPU: 40 RDNA 3.5 compute units running up to 2.9GHz, for graphics and compute-shader throughput alongside the AI silicon.
• Up to 126 TOPS combined AI performance: The XDNA 2 NPU delivers 50 TOPS on its own; 126 TOPS is the combined CPU + GPU + NPU total AMD publishes.
• 128GB unified LPDDR5x-8000 at 256GB/s: One memory pool shared across CPU, GPU and NPU, with AMD Variable Graphics Memory letting you allocate most of it to the GPU as VRAM.
• 2TB NVMe SSD pre-installed: Local storage for model weights, datasets and containers.
• Full ROCm support out of the box: AMD's ROCm stack comes pre-configured, with AMD AI Playbooks and ready-to-run setups for LM Studio, ComfyUI and VS Code.
• 10GbE, Wi-Fi 7 and Bluetooth 5.4: Wired and wireless networking, plus 3x USB-C ports (with DisplayPort Alt Mode) and HDMI 2.1b.
• Compact chassis: 150 x 150 x 45.4mm, roughly 1.2kg, with a 120W TDP.

Specifications

Good to know

• ROCm has matured substantially, but it does not yet match CUDA's pre-built container images, third-party tutorials and community troubleshooting history. For teams invested in CUDA-specific tooling, that is a real migration cost.
• It ships as AMD reference hardware with a custom AMD BIOS, and you choose either Windows 11 Pro or a Linux configuration.

Choose the Ryzen AI Halo Developer Platform if you need x86 compatibility and a well-supported ROCm setup for local AI development. If your workflow is CUDA-first, an NVIDIA GB10 system is the closer match.