AMD is making a modest update to the Ryzen AI MAX product line, this time codenamed Gorgon Halo, which is set to replace the existing Ryzen AI MAX 300 "Strix Halo." While the core specifications remain largely unchanged, the adjustments are primarily focused on frequency, voltage profile, and platform support, following the typical "same mold, same frequency" route.

Based on leaked information, the Ryzen AI MAX 400 series is comprised of five major SKUs, covering configurations ranging from 8-core to 16-core. The CPU architecture continues to be Zen 5, while the GPU maintains its RDNA 3.5 structure. Notably, both the CPU and GPU's maximum operating frequencies have been increased, alongside an upgrade to the memory subsystem, which now supports LPDDR5X-8533, offering higher bandwidth compared to the previous LPDDR5X-8000 supported by Strix Halo. This enhancement is particularly favorable for sustained loads involving integrated GPUs and NPUs, especially in UMA architectures where memory bandwidth often becomes a bottleneck before the number of compute units does.

The flagship model, Ryzen AI MAX+ 495, features a 16-core, 32-thread configuration along with a full-spec Radeon 8060S GPU encompassing 40 compute units. The CPU's base and boost frequencies have been raised to 3.1GHz and 5.2GHz, respectively, with the GPU frequency also increased to 3.0GHz, all by 100MHz. These are not drastic increments but more of a redistribution of the optimally manufactured wafers within the current voltage and cooling constraints. The chip's nominal TDP remains at 55W, although it can be lowered to 45W or elevated to 120W by OEMs under suitable thermal conditions, offering greater design flexibility for high-performance ultrabooks or compact workstations.

Intermediate models like the Ryzen AI MAX+ 492 and MAX 490 reduce the core count to 12 cores and 24 threads. The differentiation between them lies in the GPU units, with the former retaining 40 CU 8060S and the latter housing 32 CU 8050S. Additionally, the 8-core models, MAX+ 488 and MAX 485, follow a similar principle of tier differentiation based on GPU specifications. This segmentation continues the Strix Halo's logic of defining price bands through GPU castration rather than redesigning chips of varying sizes.

It's important to note that Gorgon Halo maintains a flat 55W design starting point, highlighting that AMD is not increasing default power consumption for performance gains. Instead, it builds incremental performance improvements on frequency tuning and memory bandwidth enhancements, a strategic move that is generally more acceptable to OEMs on mobile platforms and contributes to better noise control and endurance metrics.

Regarding the NPU, no official details have been released yet, but given the frequency tuning and memory upgrades, it is speculated that its processing capability will fall within the 55-60 TOPS range, representing a modest progression from the existing Ryzen AI MAX 300. This level of enhancement is more about aligning with platform marketing and software adaptation, rather than achieving qualitative algorithmic changes.

In terms of product evolution, Gorgon Halo's introduction represents more of a platform stabilization move. The continued replenishment of SKUs related to Strix Halo in the near term indicates that AMD is not rushing a full generation switch. Instead, it seeks to extend the lifecycle of the existing design by iterating on frequency and memory support. With the standard Ryzen AI 400 "Gorgon" poised to enter the market soon, AMD aims to cover both high-performance APUs and mainstream AI PCs on the mobile side, in preparation for the upcoming Intel Core Ultra Series 3 "Panther Lake." Against this backdrop, Gorgon Halo's mission is straightforward: to maintain a performance edge without introducing new architectural risks, offering OEMs a more refined high-end integration solution.