← All hardware
The most powerful x86 APU, pairing 16 Zen 5 cores with up to 128GB unified memory for local LLMs.
Pros
- Best performance-per-dollar for local LLMs at its tier
- 128GB unified memory in mini-PC/laptop form factors
- Strong iGPU rivals entry discrete cards
Cons
- NPU TOPS (50) below Snapdragon X2
- No CUDA; ROCm/Vulkan ecosystem less mature
- High power draw under full load vs Arm chips; 128GB prices spiking in 2026
✓ Where it shines / best for
- Local LLM inference and AI development needing large unified memory
- Compact AI workstations and high-end mini-PCs
- Creators running GPU-accelerated workloads without a discrete GPU
✕ Not the best fit for
- Mainstream thin-and-light laptops (high power/cost)
- Budget buyers
- Users who only need basic Copilot+ AI features
Features
- ✓ On-device / offline
- ✓ NPU
- ✓ Unified Memory
- ✓ Local LLM
- ✓ 50 Tops
- ✓ Integrated Graphics
- ✓ Xdna2
- ✓ Zen5
- ✓ AI Workstation
Pricing
| Plan | Price | Billing | Notes |
|---|---|---|---|
| OEM / embedded in device | ~$1,999+ | one-time | Sold inside premium laptops, mini-PCs and workstations (e.g. Asus ROG Flow Z13, HP ZBook Ultra, Framework Desktop, GMKtec EVO-X2). 128GB unified-memory configs typically run $2,000-$3,500+. |
Pricing verified from the official source. Prices change often — confirm on the vendor's site before buying.
Specifications
| gpu | 40-CU RDNA 3.5 (Radeon 8060S) |
| npu | XDNA 2, 50+ TOPS |
| tdp | 45-120W configurable |
| memory | Up to 128GB unified LPDDR5X (96GB as VRAM) |
| cpu_cores | 16 Zen 5 / 32 threads |
| max_clock | 5.1 GHz |
| architecture | Strix Halo, TSMC 4nm APU |
Sponsored
A full review is being generated for this product and will appear here shortly.