Can I Run Nemotron 3 Super on Framework Desktop (Ryzen AI Max+ 395, 128 GB)?

Autor: Jakub Rusinowski · Ostatnia aktualizacja: 15 sierpnia 2026

Yes

Yes — Nemotron 3 Super 120B-A12B at Q6_K needs about 102.1 GB of the 128 GB usable on Framework Desktop (Ryzen AI Max+ 395, 128 GB) (~25.9 GB spare), at ~16.4 tok/s (estimated), with room for about 65,536 tokens of context.

Confidence: medium · Recommended quantization: Q6_K · Estimated speed: ~16.4 tok/s

Buy This HardwareApple Mac Studio M4 Max — 128 GB VRAM · 35 W board power Deploy in the Cloud NowRunPod

or compare on Vast.ai

As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.

Framework Desktop (Ryzen AI Max+ 395, 128 GB) — what it gives a model

Usable memory for models	128 GB
Memory bandwidth	256 GB/s
Form factor	Mini PC
Operating system	Windows or Linux
Memory upgradeable	No — soldered

Nemotron 3 Super on Framework Desktop (Ryzen AI Max+ 395, 128 GB): memory by quantization

Quant	Memory needed	Fits 128 GB?	Max context	Est. speed	Download
F16	243.7 GB	✗ No	—	—	240 GB
Q8_0	131.2 GB	✗ No	—	—	127.5 GB
Q6_K	102.1 GB	✓ Yes	64K	~16.4 tok/s	98.4 GB
Q5_K_M	88.7 GB	✓ Yes	64K	~18.5 tok/s	85.1 GB
Q4_K_M	76.1 GB	✓ Yes	128K	~21.1 tok/s	72.5 GB
Q3_K_M	54.8 GB	✓ Yes	128K	~27.5 tok/s	51.2 GB
Q2_K	43.1 GB	✓ Yes	128K	~33.1 tok/s	39.4 GB

What to watch out for

This model does not publish its full attention configuration, so the KV-cache share of these figures is inferred from its parameter count rather than computed exactly.
Memory on this machine is not upgradeable, so the configuration you buy is the ceiling for every model you will ever run on it.

Framework Desktop 128 GB limitations

Memory is soldered LPDDR5X — unusually for Framework, this is the one component that cannot be upgraded.
ROCm/Vulkan support for Strix Halo is younger than CUDA; check your runtime supports it before buying.

Recommended setup

Ollama or llama.cpp (CUDA/ROCm) — vLLM if you need concurrent serving

How these numbers are calculated

Memory need = quantized weights + KV cache at 8,192 tokens (f16) + 0.8 GB runtime overhead.
128 GB unified memory at 256 GB/s, shared between CPU and GPU.
Throughput is a memory-bandwidth roofline estimate, not a measurement. It is labelled "estimated" everywhere it appears.
KV cache is inferred from the parameter count because this model does not publish its full attention configuration.

FAQ