AMD Ryzen AI Max+ 395 (Strix Halo)
AMD's first true Strix Halo chip, combining Zen 5 CPU cores, RDNA 3.5 integrated graphics, and a 50-TOPS XDNA 2 NPU on a single die. The headline feature for local AI is **up to 128 GB of unified LPDDR5X-8000 memory** that the GPU tile can address directly — meaningful for 70B-class models that don't fit on any consumer discrete GPU.
Extrapolated from 256 GB/s bandwidth — 25.6 tok/s estimated. No measured benchmarks yet.
Plain-English: Doesn't fit modern chat models usefully — vision models won't fit.
Verdicts extrapolated from catalog VRAM + bandwidth + ecosystem flags. Hover any chip for the rationale. Want measured numbers? Submit your own run with runlocalai-bench --submit.
Overview
AMD's first true Strix Halo chip, combining Zen 5 CPU cores, RDNA 3.5 integrated graphics, and a 50-TOPS XDNA 2 NPU on a single die. The headline feature for local AI is **up to 128 GB of unified LPDDR5X-8000 memory** that the GPU tile can address directly — meaningful for 70B-class models that don't fit on any consumer discrete GPU. Memory bandwidth (~256 GB/s) sits well below a discrete card's GDDR6X / GDDR7 (which run 600-1500 GB/s), so tokens-per-second on large models is slower than dense-card alternatives. The trade-off operators are making: capacity over speed. Apple's M3/M4 Max chips compete in the same niche. Ships in mini-PCs and laptops from multiple OEMs through 2025-2026.
Search-fallback link — editorial hasn't yet curated a retailer URL for this card. Approx. $2300.
Some links above are affiliate links. We may earn a commission at no extra cost to you. How we make money.
Specs
| System RAM (typical) | 64 GB |
| Power draw | 120 W |
| Released | 2025 |
| Backends | ROCm Vulkan |
Models that fit
Open-weight models small enough to run on AMD Ryzen AI Max+ 395 (Strix Halo) with usable context.
Frequently asked
Does AMD Ryzen AI Max+ 395 (Strix Halo) support CUDA?
How much does AMD Ryzen AI Max+ 395 (Strix Halo) cost?
Where next?
Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify hardware specifications.