Which local LLM can be run on Ryzen AI Max+ 395 128GB?

Compare K2 Horizon 0.9B, Qwen3.8-Flash-Next with a 4K input and the recommended format for each hardware platform. Required memory varies by model and context length.

How much power does the local LLM of Ryzen AI Max+ 395 with 128GB consume?

Expected power consumption under sustained load is approximately 125W, with an actual range anticipated between 80–170W.

Are the displayed token speeds ground truth?

These ranges normalize collected benchmark measurements to the same conditions. Only combinations without matching measurements are estimated from nearby measured values. Results can vary by runtime, cooling, model file, and context length.

Updated September 2026

Ryzen AI Max+ 395 128GB This is a choice aimed at loading a 128GB model at a low body price.

It is attractive when you have large amounts of integrated memory at a low price and can handle runtime settings directly.

Speed and memory for the same model

Ryzen AI Max+ 395 128GB · Qwen3.8 27B · local LLM speed and memory: Prompt processing · prefill 48–88 tok/s · Token generation · decode 11.2–12.5 tok/s; Required 17.56GB / Usable 96GB unified memory — 4,096 input tokens · single request · prompt cache off

Calculate speed with this configuration Method and evidence CC BY 4.0

Available memory: 96GB
Memory bandwidth: 256GB/s
Estimated sustained LLM load: About 125W
Korean price range (KRW): 538 × KRW 10,000–562 × KRW 10,000

Who it fits

Loading large models
Select Windows/Linux
128GB cost-effectiveness

Skip it if

Even if there is ample memory, the actual speed difference is large depending on the runtime and GPU acceleration support, so you should check compatibility for each model first.

Recommended format for each device 4K input

The model you will experience with this equipment

Basic decoding · Expected range

K2 Horizon 0.9B

Required 1.03GB · spare time

llama.cpp HIP · Q4_K_M

Decode

338.1 ~ 376.8 tok/s

First token

1.6 to 3.1 seconds

Qwen3.8-Flash-Next

recommended

Required 57.8561152GB · spare time

Strata 0.1.40 · IQ3_S · ROCm 7.14.1 · MTP-4

Decode

44.9 ~ 73.3 tok/s

First token

2.6 to 4.3 seconds

speed tuning

Actual speed varies with runtime version, cooling, context length and quantization file.

purchase judgment

What tasks is it suitable for?

Large LLMs

Suitable

We can review the larger Q4 model with 112GB of available memory.

Document/RAG

Suitable

It has a large capacity to hold both the model and long context.

Convenience of installation

Conditional

You should check runtime support for each model yourself.

The first thing to get stuck: GPU acceleration support by runtime and kernel completeness determine actual usability rather than hardware capacity.
Configuration to check: Ryzen AI Max+ 395 · System memory 128GB · NVMe 2TB recommended
Criteria for spending more money: It is advantageous if you want to put up with the settings and install a large capacity model cheaply. If you want reliable tools out of the box, you're better off with a Mac or NVIDIA finished product.

On sales pages, look at memory before chips.

The model and speed assumptions above change if the product is not the 128GB configuration.

Frequently asked questions

Which local LLM can be run on Ryzen AI Max+ 395 128GB?: K2 Horizon 0.9B, Qwen3.8-Flash-Next and others with a 4K input and the recommended format. Required memory varies by model and context length.
How much power does the local LLM of Ryzen AI Max+ 395 with 128GB consume?: Sustained load is about 125 W, with an actual range of 80–170 W.
Are the displayed token speeds ground truth?: This is a range aligned to collected benchmark measurements under the same conditions. Combinations without a matching measurement are estimated from the nearest result and vary with runtime, cooling, quantization file and context length.

MLX completeness

Studio M4 Max 128GB

Suits people who want to compare a wide range from 70B-class to high-precision models in a single finished machine.

CUDA environment

DGX Spark 128GB

This is for handling a large model on one machine without splitting it across several graphics cards. You should have a reason to buy 128GB. If the model you need now fits in 24GB or 32GB, do not assume Spark will be faster just because it has more memory.

Single GPU speed

RTX 5090 32GB

Compare this card if you repeatedly generate images or run 27B–35B inference and want shorter waits. But 32GB is still a limit. Buying on compute performance alone, without checking the model, input and image resolution, may leave you offloading again.