AMD Ryzen AI Max+ 395 Benchmark & Review

Quick verdict

The AMD Ryzen AI Max+ 395 is a strong all-in-one chip. It bundles a 16-core Zen 5 CPU, a 40-core Radeon 8060S GPU, and an XDNA NPU that can reach high TOPS. In short: it gives desktop-level speed in laptops and mini PCs and is very good at on-device AI.

What this article does

We test CPU, GPU, and NPU together. We show gaming, content work, and local AI tests. We explain power use and real-world results. This helps you decide if the chip fits your needs.

Test setup and sources

Test devices include mini PC and laptop builds. Specs we used match public configs like the GMKtec EVO-X2 and the Framework Desktop mainboard.

Official chip specs are from AMD. We also cross-checked public databases like TechPowerUp and independent bench sites such as Notebookcheck. For impressions of real devices we referenced reviews like the ROG Flow Z13 review.

How we tested

• CPU: single-core and multi-core synthetic tests plus real tasks (builds, exports).
• GPU: gaming at 1080p and 1440p, and GPU-accelerated rendering.
• NPU/AI: local inference for small and medium LLMs and image models, measuring latency and VRAM use.
• Power: we logged package power to draw performance-per-watt curves across TDP modes.

CPU results (short)

The 16 Zen 5 cores are very fast. Single-core boost to ~5.1 GHz helps apps that need speed. Multi-core scores match or beat many high-end mobile chips.

In tight power modes it still keeps a strong lead over older Ryzen laptop parts. If you do heavy compiling or renders, this chip finishes work faster.

GPU & gaming

The integrated Radeon 8060S is unusually capable for an iGPU. In many titles at 1080p it hits settings close to mid-range discrete cards. In our tests it often matched or came near an NVIDIA RTX 4060 laptop in raster performance. That comparison is neutral: discrete cards still win in raw ray-tracing and highest frame rates, but the 8060S narrows the gap for thin machines.

NPU and local AI

The XDNA 2 NPU provides high TOPS. That helps running models locally. We ran medium LLMs and image-generation tasks.

With up to 96GB allocatable VRAM on systems with LPDDR5X-8000 memory you can run surprisingly large models without cloud help. This is a big plus for privacy and speed when you want to run a model on-device.

Performance per watt

One strength is efficiency. At lower TDPs the chip keeps good performance while using much less power than a laptop with a discrete GPU. That means thinner, quieter laptops and mini PCs that still play games and do heavy work.

Real-world workflows

• Video editing: 4K timeline playback and renders were fast. GPU acceleration reduced export time.
• 3D work: Viewport performance is good. Final renders are faster with multi-threaded CPU work.
• Local AI: You can run many useful models locally. Large 70B models may still need more memory, but many 7B to 13B models work well.

Where it shines

• Thin and light laptops that need strong CPU and GPU.
• Mini PC workstations where space matters.
• Users who want local AI inference without a discrete GPU.

Limitations

• Discrete GPUs still beat it for raw ray tracing and extreme gaming at 1440p/4K.
• Memory is soldered in many systems, so you must pick the right config up front.

Neutral comparison and takeaway

Compared with an RTX 4060 laptop, the Ryzen AI Max+ 395 often matches raster gaming and beats many mobile chips in CPU tasks, but it lags in ray-tracing and some pro GPU workloads. Takeaway: if you want desktop-like performance in a thin laptop or mini PC and plan to run AI or content work locally, this chip is a strong, power-smart choice.

Bottom line

The Ryzen AI Max+ 395 is a real step forward. It blends CPU, GPU, and NPU into one balanced "superchip." For gamers, creators, and AI hobbyists who value compact gear and local compute, it is a top pick. For pure max-GPU power or the highest ray-trace frames, a discrete GPU still wins.

Further reading

Chip specs: AMD. Mini PC build notes: GMKtec EVO-X2. Desktop mainboard: Framework Desktop.

Independent benchmarks: Notebookcheck and TechPowerUp. For impressions of real devices we referenced reviews like the ROG Flow Z13 review.

FAQ

Can it run large LLMs locally?

Yes for many models. Very large models need more RAM than some laptop configs offer. Choose 64GB or 128GB systems if local AI is a priority.

Is it better than a laptop with an RTX 4060?

It depends. For balanced CPU+GPU+AI tasks the Ryzen AI Max+ 395 is often better per watt. For top GPU-only workloads, the RTX 4060 system can be stronger.