Text model · Nemotron 3

NE Nemotron 3 Super 120B-A12B: RAM and VRAM requirements

Nemotron 3 Super 120B-A12B needs about 80.3 GB to run at Q4_K_M (the Q4_K_M GGUF file is ~76.87 GB to download; KV cache and overhead add the rest), or about 123.1 GB at Q8_0. The lightest hardware that runs it is Apple M4 Max (128GB).

Nemotron 3 family · 120B params (Mixture-of-Experts: activates only 12B of 120B params per token, so generation is faster than the total size suggests) · released Mar 2026.

Shopping for hardware? See what runs Nemotron 3 Super 120B-A12B →

License nvidia-nemotron-open-model-license · Conditional ↓ 875.4K/mo ♥ 414 on HuggingFace

Q4_K_M GGUF

76.87 GB

Q8_0 GGUF

119.65 GB

Memory @ Q4 (4k)

~80.3 GB

Context

256 k

Q4_K_M and Q8_0 are GGUF download (file) sizes. Memory @ Q4 includes the KV cache and overhead, so it runs larger than the file on disk.

Will it run on your device?

Nemotron 3 Super 120B-A12B runs on 4 of 40 tracked devices at Q4_K_M.

4 run well 0 tight fit 36 too small

Fit check Q4_K_M

Your device

Apple M4 (16GB)

No, not enough memory

needs 80.3 GB usable 10.5 GB

See the full breakdown

Memory breakdown

Weights (Q4_K_M)76.87 GB

KV cache (4k)2.6 GB

Overhead0.8 GB

Total80.3 GB

How context length changes it

4k context ~80.3 GB 32k context ~98.7 GB 128k context ~161.8 GB

Longer context grows the KV cache, which for Nemotron 3 Super 120B-A12B is sized by its 12B active params, not the full 120B. It needs ~80.3 GB at 4k and ~161.8 GB at 128k.

Quantization sizes

GGUF quantson disk

Quantization	Size on disk
Q2_K	50.3 GB est
Q3_K_M	58.7 GB est
Q4_K_M (default)	76.87 GB
Q5_K_M	85.5 GB est
Q6_K	98.4 GB est
Q8_0	119.65 GB
FP16	224.93 GB

Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.

Run it

Ollama

$ ollama run nemotron-3-super:120b-a12b

llama.cpp

$ llama-cli -hf unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF:Q4_K_M

LM Studio

$ lms get unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF

Which devices can run Nemotron 3 Super 120B-A12B?

Apple Silicon Macs

RAM-only laptops

iPhone & iPad

Android

NVIDIA GPUs

AMD GPUs

Same job, different size

Pick by what fits your memory: step down to free up VRAM, or step up if you have headroom.

Next size down Laguna S 2.1 118B Next size up dots.llm1 142B

Similar models

Head-to-head

FAQ

How much VRAM or RAM does Nemotron 3 Super 120B-A12B need?

At Q4_K_M, Nemotron 3 Super 120B-A12B needs about 80.3 GB (weights ~76.87 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~123.1 GB.

What is the Q4_K_M GGUF file size of Nemotron 3 Super 120B-A12B?

The Q4_K_M GGUF file is about 76.87 GB to download, and the Q8_0 GGUF is about 119.65 GB. That is the weights file on disk; to run it you also need room for the KV cache and overhead, so budget ~80.3 GB of memory at Q4_K_M.

Can Nemotron 3 Super 120B-A12B run on a laptop?

Nemotron 3 Super 120B-A12B is large; you need a high-memory Mac or multi-GPU setup at Q4_K_M.

Is Nemotron 3 Super 120B-A12B cheaper to run because it is a MoE model?

It is faster, not lighter. Nemotron 3 Super 120B-A12B activates only 12B of 120B params per token (so it runs quickly), but all experts must stay in memory, so it still needs memory for the full 120B.

Can I use Nemotron 3 Super 120B-A12B commercially?

Conditionally. NVIDIA Nemotron Open Model License; check the terms for commercial use.

Understand the numbers

Short guides to the ideas behind Nemotron 3 Super 120B-A12B's memory and quant figures.

MoE: 120B total / 12B active, hybrid Latent-MoE (Mamba-2 + MoE + multi-token-prediction layers), NVFP4-trained. 256K context by default, up to 1M. Needs a multi-GPU rig or a high-memory Mac Studio even at Q4. Sizes from unsloth's UD-quant GGUF repo. NVIDIA Nemotron Open Model License.

Sources

Last validated 2026-08-03. Memory figures are estimates. See methodology.