
AMD
AI · Hardware
AMD debuts Ryzen AI Max 400 ‘Gorgon Halo’ with up to 192GB unified memory
AMD’s new Ryzen AI Max PRO 400 ‘Gorgon Halo’ APUs bring up to 192GB unified memory and 55 TOPS NPUs for AI-heavy commercial PCs and agentic workloads, shipping from Q3 2026.
AMD has introduced the Ryzen AI Max PRO 400 Series, codenamed Gorgon Halo, a refreshed family of APUs built on its Strix Halo silicon that targets AI-intensive professional systems. The chips combine up to 16 Zen 5 CPU cores, RDNA 3.5 integrated graphics, and an XDNA 2 NPU in a single SoC, and raise the maximum unified LPDDR5X memory capacity from 128 GB in the prior 300 series to 192 GB, of which up to 160 GB can function as GPU-accessible VRAM. The flagship Ryzen AI Max+ PRO 495 features 16 cores and 32 threads, a 5.2 GHz boost clock, 80 MB of total cache, a Radeon 8065S iGPU with 40 compute units, and an NPU rated up to 55 TOPS, within a configurable 45–120 W TDP envelope.
The Ryzen AI Max PRO 400 lineup currently consists of three SKUs—the Max+ PRO 495, Max PRO 490, and Max PRO 485—all branded with AMD PRO technologies and positioned for commercial, workstation-class and AI developer use rather than mainstream consumer laptops. According to AMD, these processors are designed to run local large language models with more than 300 billion parameters and support complex, concurrent agent-based AI workflows, effectively turning client systems into "agent computers" capable of advanced planning and task execution without cloud reliance. OEM systems using the Gorgon Halo chips, including platforms from HP and Lenovo and small-form-factor desktops such as AMD’s own Ryzen AI Halo, are slated to roll out starting in the third quarter of 2026, with AMD confirming global regional availability for both laptop and desktop form factors. Initial Ryzen AI Halo developer boxes launch first with 300-series silicon and up to 128 GB unified memory, with a next-generation Halo platform based on the Ryzen AI Max PRO 400 Series and up to 192 GB unified memory following later in the same timeframe.