Skip to content

NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 VRAM requirements

How much VRAM NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 needs
4-bit weights: 70.3 GBWith 8K context: 71.3 GBWith 32K context: 71.5 GB8-bit, 8K context: 124 GB
Parameters
124B
Active per token
12B
Longest context
1M
Released
Oct 2026
License
Openmdw 1.1
Strongest at
Text and images

Its licence (Openmdw 1.1) is not a standard open-source one: check the terms before commercial use.

How much VRAM NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 needs

Memory NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit232 GB233 GB233 GB234 GB
8-bit123 GB124 GB124 GB125 GB
6-bit95.0 GB96.1 GB96.3 GB97.0 GB
5-bit82.4 GB83.5 GB83.7 GB84.4 GB
4-bit70.3 GB71.3 GB71.5 GB72.3 GB
3-bit56.6 GB57.7 GB57.9 GB58.6 GB
2-bit43.5 GB44.5 GB44.7 GB45.4 GB
1.6-bit23.9 GB25.0 GB25.1 GB25.9 GB

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

GPUs and Macs that run NVIDIA Nemotron 3.5 Super VL 120B A12B BF16

The best way NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 4060Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 4060 report
RTX 5060Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 5060 report
RTX 3070Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 3070 report
RTX 3080 10 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 3080 10 GB report
RTX 3060 12 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 3060 12 GB report
Arc B580Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Arc B580 report
RTX 4070Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 4070 report
RTX 4070 SuperToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 4070 Super report
RTX 5070Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 5070 report
RTX 4060 Ti 16 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 5060 Ti 16 GB report
Arc A770 16 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Arc A770 16 GB report
Radeon RX 9070 XTToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Radeon RX 9070 XT report
RTX 4070 Ti SuperToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 4070 Ti Super report
RTX 4080 SuperToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 4080 Super report
RTX 5070 TiToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 5070 Ti report
RTX 5080Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 5080 report
Radeon RX 7900 XTXToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Radeon RX 7900 XTX report
RTX 3090Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 3090 report
RTX 4090Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 4090 report
Radeon AI PRO R9700Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Radeon AI PRO R9700 report
RTX 5090Too bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 5090 report
2× RTX 30902-bit64K81 tok/s1.3 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on 2× RTX 3090 report
2× RTX 40902-bit64K84 tok/s0.6 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB2-bit144K98 tok/s1.2 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada2-bit144K82 tok/s0.9 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX 6000 Ada report
L40S2-bit144K77 tok/s1.5 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on L40S report
2× RTX 50903-bit320K99 tok/s0.5 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on 2× RTX 5090 report
4× RTX 30905-bit736K54 tok/s0.7 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on 4× RTX 3090 report
RTX PRO 6000 Blackwell5-bit992K82 tok/s0.6 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX PRO 6000 Blackwell report
Mac mini M4 16 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Mac mini M4 16 GB report
Mac mini M4 24 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Mac mini M4 24 GB report
Mac mini M4 32 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GBToo bigPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB2-bit168K30 tok/s27 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Mac mini M4 Pro 64 GB report
M4 Max 128 GB5-bit992K30 tok/s13 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on M4 Max 128 GB report
M5 Max 128 GB5-bit992K32 tok/s3.9 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on M5 Max 128 GB report
M3 Ultra 96 GB3-bit1M42 tok/s7.5 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit1M27 tok/s7.5 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit1M27 tok/s7.5 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit1M35 tok/s2.3 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB2-bit168K26 tok/s7.7 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB5-bit992K15 tok/s7.7 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Ryzen AI Max+ 395 128 GB report
RTX Spark6-bit1M19 tok/s2.7 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on RTX Spark report
DGX Spark6-bit1M17 tok/s2.7 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on DGX Spark report
Dual DGX Spark8-bit1M14 tok/s1.6 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on Dual DGX Spark report
DGX Station GB3008-bit1M127 tok/s0.2 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit1M64 tok/s0.4 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit1M64 tok/s0.2 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit1M64 tok/s0.1 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM4-bit576K95 tok/s1.7 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on A100 80 GB SXM report
H100 SXM4-bit576K118 tok/s0.5 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on H100 SXM report
4× H200 NVL8-bit1M110 tok/s0.2 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 on 4× H200 NVL report
8× H200 NVL8-bit1M110 tok/s0.1 sPDF: NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 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

llama.cpp
llama-server -hf ggml-org/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-GGUF
vLLM
vllm serve nvidia/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16

Questions

How much VRAM does NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 need?

NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 needs 70.3 GB for its weights at 4-bit, 71.3 GB with an 8K context and 71.5 GB with 32K. At 8-bit, 124 GB.

What is the minimum VRAM for NVIDIA Nemotron 3.5 Super VL 120B A12B BF16?

About 44.5 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 NVIDIA Nemotron 3.5 Super VL 120B A12B BF16?

124B parameters (12B active per token): 232 GB of weights at 16-bit, 70.3 GB at 4-bit.

Can NVIDIA Nemotron 3.5 Super VL 120B A12B BF16 run on an RTX 4090?

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

More models