Skip to content

Nemotron 3 Super 120B A12B VRAM requirements

How much VRAM Nemotron 3 Super 120B A12B needs
4-bit weights: 69.8 GBWith 8K context: 70.8 GBWith 32K context: 71.0 GB8-bit, 8K context: 123 GB
Parameters
124B
Active per token
12B
Longest context
256K
Released
Mar 2026
License
NVIDIA Nemotron Open Model
Strongest at
Chat, Coding, Agents, Docs

How much VRAM Nemotron 3 Super 120B A12B needs

Memory Nemotron 3 Super 120B A12B needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit230 GB231 GB231 GB232 GB
8-bit122 GB123 GB124 GB124 GB
6-bit94.4 GB95.4 GB95.6 GB96.4 GB
5-bit81.9 GB82.9 GB83.1 GB83.9 GB
4-bit69.8 GB70.8 GB71.0 GB71.8 GB
3-bit56.3 GB57.3 GB57.5 GB58.2 GB
2-bit43.2 GB44.2 GB44.4 GB45.2 GB
1.6-bit23.7 GB24.8 GB25.0 GB25.7 GB

One conversation, 16-bit context cache, runtime overhead included.

GPUs and Macs that run Nemotron 3 Super 120B A12B

The best way Nemotron 3 Super 120B A12B runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: Nemotron 3 Super 120B A12B on RTX 4060 report
RTX 5060Too bigPDF: Nemotron 3 Super 120B A12B on RTX 5060 report
RTX 3070Too bigPDF: Nemotron 3 Super 120B A12B on RTX 3070 report
RTX 3080 10 GBToo bigPDF: Nemotron 3 Super 120B A12B on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: Nemotron 3 Super 120B A12B on RTX 3060 12 GB report
Arc B580Too bigPDF: Nemotron 3 Super 120B A12B on Arc B580 report
RTX 4070Too bigPDF: Nemotron 3 Super 120B A12B on RTX 4070 report
RTX 4070 SuperToo bigPDF: Nemotron 3 Super 120B A12B on RTX 4070 Super report
RTX 5070Too bigPDF: Nemotron 3 Super 120B A12B on RTX 5070 report
RTX 4060 Ti 16 GBToo bigPDF: Nemotron 3 Super 120B A12B on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GBToo bigPDF: Nemotron 3 Super 120B A12B on RTX 5060 Ti 16 GB report
Arc A770 16 GBToo bigPDF: Nemotron 3 Super 120B A12B on Arc A770 16 GB report
Radeon RX 9070 XTToo bigPDF: Nemotron 3 Super 120B A12B on Radeon RX 9070 XT report
RTX 4070 Ti SuperToo bigPDF: Nemotron 3 Super 120B A12B on RTX 4070 Ti Super report
RTX 4080 SuperToo bigPDF: Nemotron 3 Super 120B A12B on RTX 4080 Super report
RTX 5070 TiToo bigPDF: Nemotron 3 Super 120B A12B on RTX 5070 Ti report
RTX 5080Too bigPDF: Nemotron 3 Super 120B A12B on RTX 5080 report
Radeon RX 7900 XTXToo bigPDF: Nemotron 3 Super 120B A12B on Radeon RX 7900 XTX report
RTX 3090Too bigPDF: Nemotron 3 Super 120B A12B on RTX 3090 report
RTX 4090Too bigPDF: Nemotron 3 Super 120B A12B on RTX 4090 report
Radeon AI PRO R9700Too bigPDF: Nemotron 3 Super 120B A12B on Radeon AI PRO R9700 report
RTX 5090Too bigPDF: Nemotron 3 Super 120B A12B on RTX 5090 report
2× RTX 30902-bit96K81 tok/s1.3 sPDF: Nemotron 3 Super 120B A12B on 2× RTX 3090 report
2× RTX 40902-bit96K84 tok/s0.6 sPDF: Nemotron 3 Super 120B A12B on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB2-bit184K98 tok/s1.2 sPDF: Nemotron 3 Super 120B A12B on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada2-bit184K82 tok/s0.9 sPDF: Nemotron 3 Super 120B A12B on RTX 6000 Ada report
L40S2-bit184K77 tok/s1.5 sPDF: Nemotron 3 Super 120B A12B on L40S report
2× RTX 50903-bit256K99 tok/s0.5 sPDF: Nemotron 3 Super 120B A12B on 2× RTX 5090 report
4× RTX 30905-bit256K54 tok/s0.7 sPDF: Nemotron 3 Super 120B A12B on 4× RTX 3090 report
RTX PRO 6000 Blackwell5-bit256K82 tok/s0.6 sPDF: Nemotron 3 Super 120B A12B on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: Nemotron 3 Super 120B A12B on Mac mini M4 16 GB report
Mac mini M4 24 GBToo bigPDF: Nemotron 3 Super 120B A12B on Mac mini M4 24 GB report
Mac mini M4 32 GBToo bigPDF: Nemotron 3 Super 120B A12B on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GBToo bigPDF: Nemotron 3 Super 120B A12B on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GBToo bigPDF: Nemotron 3 Super 120B A12B on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB2-bit208K30 tok/s27 sPDF: Nemotron 3 Super 120B A12B on Mac mini M4 Pro 64 GB report
M4 Max 128 GB5-bit256K30 tok/s13 sPDF: Nemotron 3 Super 120B A12B on M4 Max 128 GB report
M5 Max 128 GB5-bit256K32 tok/s3.9 sPDF: Nemotron 3 Super 120B A12B on M5 Max 128 GB report
M3 Ultra 96 GB3-bit256K42 tok/s7.5 sPDF: Nemotron 3 Super 120B A12B on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit256K27 tok/s7.5 sPDF: Nemotron 3 Super 120B A12B on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit256K27 tok/s7.5 sPDF: Nemotron 3 Super 120B A12B on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit256K35 tok/s2.3 sPDF: Nemotron 3 Super 120B A12B on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB2-bit208K26 tok/s7.7 sPDF: Nemotron 3 Super 120B A12B on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB5-bit256K15 tok/s7.7 sPDF: Nemotron 3 Super 120B A12B on Ryzen AI Max+ 395 128 GB report
RTX Spark6-bit256K19 tok/s2.7 sPDF: Nemotron 3 Super 120B A12B on RTX Spark report
DGX Spark6-bit256K17 tok/s2.7 sPDF: Nemotron 3 Super 120B A12B on DGX Spark report
Dual DGX Spark8-bit256K14 tok/s1.6 sPDF: Nemotron 3 Super 120B A12B on Dual DGX Spark report
DGX Station GB3008-bit256K127 tok/s0.2 sPDF: Nemotron 3 Super 120B A12B on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit256K64 tok/s0.4 sPDF: Nemotron 3 Super 120B A12B on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit256K64 tok/s0.2 sPDF: Nemotron 3 Super 120B A12B on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit256K64 tok/s0.1 sPDF: Nemotron 3 Super 120B A12B on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM4-bit256K95 tok/s1.7 sPDF: Nemotron 3 Super 120B A12B on A100 80 GB SXM report
H100 SXM4-bit256K118 tok/s0.5 sPDF: Nemotron 3 Super 120B A12B on H100 SXM report
4× H200 NVL8-bit256K110 tok/s0.2 sPDF: Nemotron 3 Super 120B A12B on 4× H200 NVL report
8× H200 NVL8-bit256K110 tok/s0.1 sPDF: Nemotron 3 Super 120B A12B on 8× H200 NVL report

Best fit is the highest quality, up to 8-bit, that leaves 16K of context. First reply is the wait for a 2K prompt. Speeds are rough, plus or minus 25%.

How to run it

Ollama
ollama run nemotron-3-super:120b-a12b
llama.cpp
llama-server -hf unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-GGUF
vLLM
vllm serve nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16
MLX (Mac)
mlx_lm.chat --model mlx-community/NVIDIA-Nemotron-3-Super-120B-A12B-4bit

Questions

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

Nemotron 3 Super 120B A12B needs 69.8 GB for its weights at 4-bit, 70.8 GB with an 8K context and 71.0 GB with 32K. At 8-bit, 123 GB.

What is the minimum VRAM for Nemotron 3 Super 120B A12B?

About 44.2 GB at 2-bit with an 8K context, the lowest precision we recommend for a model this size. The smallest machine here that runs it without spilling is the 2× RTX 3090.

How big is Nemotron 3 Super 120B A12B?

124B parameters (12B active per token): 230 GB of weights at 16-bit, 69.8 GB at 4-bit.

Can Nemotron 3 Super 120B A12B run on an RTX 4090?

Not on one card. The smallest machine here that runs it is the 2× RTX 3090.

More models