At CES 2026, AMD launched Ryzen AI Max+ 392 and 388 for Strix Halo laptops, packing a 40-CU RDNA 3.5 iGPU up to 60 TFLOPS, XDNA 2 NPU, and performance claims.
© RusPhotoBank
At CES 2026, AMD unveiled the Ryzen AI Max+ 392 and Max+ 388, expanding the Strix Halo family aimed at high-performance laptops and compact PCs that do without a discrete GPU.
Launched a year earlier, Strix Halo targets thin-and-light machines that still need serious compute. The platform blends a multi-core CPU, integrated Radeon graphics based on RDNA 3.5, and an XDNA 2 neural processor rated at up to 50 TOPS, all tapping a unified memory pool.
Until now, the lineup was led by the 16-core Ryzen AI Max+ 395, alongside the 12-core Max 390 and the 8-core Max 385. These APUs already power devices such as the HP ZBook Ultra G1a, the Asus ROG Flow Z13, and a number of mini-PCs.
The new Ryzen AI Max+ 392 and 388 build on those parts but arrive with a markedly stronger graphics subsystem. AMD equips both with an RDNA 3.5 integrated GPU featuring 40 compute units that, by the company’s account, can deliver up to 60 TFLOPS. On paper, that is a bold figure for an iGPU.
CPU configurations see no radical shifts. The Ryzen AI Max+ 392 retains 12 cores and 24 threads with Boost clocks up to 5 GHz, while the Max+ 388 is an 8-core, 16-thread chip. The XDNA 2 NPU is unchanged as well, remaining focused on on-device AI tasks.
At the show, AMD also shared its own benchmarks. By its data, Strix Halo–based systems outpaced Nvidia DGX Spark in AI efficiency—both in performance over time and in price-to-performance. In selected AI, multitasking, and gaming scenarios, Strix Halo also showed advantages over the latest MacBook Pro with Apple’s M5 processor. As ever, such claims call for independent testing, and the comparisons will likely draw close scrutiny.
With support from OEM partners and mini-PC makers, AMD is aiming to move Strix Halo beyond narrow niches and stake a claim in the broader market for professionals and enthusiasts.