Skip to content

Nemotron 3 Nano 4B VRAM requirements

How much VRAM Nemotron 3 Nano 4B needs
4-bit weights: 2.2 GBWith 8K context: 3.2 GBWith 32K context: 3.6 GB8-bit, 8K context: 4.9 GB
Parameters
4B
Longest context
256K
Released
Mar 2026
License
NVIDIA Nemotron Open Model
Strongest at
Docs

How much VRAM Nemotron 3 Nano 4B needs

Memory Nemotron 3 Nano 4B needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit7.4 GB8.4 GB8.8 GB10.3 GB
8-bit3.9 GB4.9 GB5.3 GB6.8 GB
6-bit3.0 GB4.0 GB4.4 GB5.9 GB
5-bit2.6 GB3.6 GB4.0 GB5.5 GB
4-bit2.2 GB3.2 GB3.6 GB5.1 GB
3-bit1.8 GB2.8 GB3.2 GB4.7 GB
2-bit1.4 GB2.4 GB2.8 GB4.3 GB
1.6-bit781 MB1.8 GB2.1 GB3.6 GB

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

GPUs and Macs that run Nemotron 3 Nano 4B

The best way Nemotron 3 Nano 4B runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 40608-bit176K46 tok/s0.5 sPDF: Nemotron 3 Nano 4B on RTX 4060 report
RTX 50608-bit176K74 tok/s0.4 sPDF: Nemotron 3 Nano 4B on RTX 5060 report
RTX 30708-bit176K74 tok/s0.4 sPDF: Nemotron 3 Nano 4B on RTX 3070 report
RTX 3080 10 GB8-bit256K117 tok/s0.2 sPDF: Nemotron 3 Nano 4B on RTX 3080 10 GB report
RTX 3060 12 GB8-bit256K60 tok/s0.6 sPDF: Nemotron 3 Nano 4B on RTX 3060 12 GB report
Arc B5808-bit256K51 tok/s0.4 sPDF: Nemotron 3 Nano 4B on Arc B580 report
RTX 40708-bit256K82 tok/s0.3 sPDF: Nemotron 3 Nano 4B on RTX 4070 report
RTX 4070 Super8-bit256K82 tok/s0.2 sPDF: Nemotron 3 Nano 4B on RTX 4070 Super report
RTX 50708-bit256K106 tok/s0.2 sPDF: Nemotron 3 Nano 4B on RTX 5070 report
RTX 4060 Ti 16 GB8-bit256K49 tok/s0.3 sPDF: Nemotron 3 Nano 4B on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GB8-bit256K74 tok/s0.3 sPDF: Nemotron 3 Nano 4B on RTX 5060 Ti 16 GB report
Arc A770 16 GB8-bit256K61 tok/s0.4 sPDF: Nemotron 3 Nano 4B on Arc A770 16 GB report
Radeon RX 9070 XT8-bit256K82 tok/s0.2 sPDF: Nemotron 3 Nano 4B on Radeon RX 9070 XT report
RTX 4070 Ti Super8-bit256K106 tok/s0.2 sPDF: Nemotron 3 Nano 4B on RTX 4070 Ti Super report
RTX 4080 Super8-bit256K114 tok/s0.1 sPDF: Nemotron 3 Nano 4B on RTX 4080 Super report
RTX 5070 Ti8-bit256K135 tok/s0.2 sPDF: Nemotron 3 Nano 4B on RTX 5070 Ti report
RTX 50808-bit256K143 tok/s0.1 sPDF: Nemotron 3 Nano 4B on RTX 5080 report
Radeon RX 7900 XTX8-bit256K114 tok/s0.3 sPDF: Nemotron 3 Nano 4B on Radeon RX 7900 XTX report
RTX 30908-bit256K140 tok/s0.2 sPDF: Nemotron 3 Nano 4B on RTX 3090 report
RTX 40908-bit256K148 tok/s0.1 sPDF: Nemotron 3 Nano 4B on RTX 4090 report
Radeon AI PRO R97008-bit256K82 tok/s0.2 sPDF: Nemotron 3 Nano 4B on Radeon AI PRO R9700 report
RTX 50908-bit256K230 tok/s0.1 sPDF: Nemotron 3 Nano 4B on RTX 5090 report
2× RTX 30908-bit256K140 tok/s0.1 sPDF: Nemotron 3 Nano 4B on 2× RTX 3090 report
2× RTX 40908-bit256K148 tok/s0.1 sPDF: Nemotron 3 Nano 4B on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit256K186 tok/s0.1 sPDF: Nemotron 3 Nano 4B on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit256K143 tok/s0.1 sPDF: Nemotron 3 Nano 4B on RTX 6000 Ada report
L40S8-bit256K131 tok/s0.1 sPDF: Nemotron 3 Nano 4B on L40S report
2× RTX 50908-bit256K230 tok/s0.1 sPDF: Nemotron 3 Nano 4B on 2× RTX 5090 report
4× RTX 30908-bit256K140 tok/s0.1 sPDF: Nemotron 3 Nano 4B on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit256K230 tok/s0.1 sPDF: Nemotron 3 Nano 4B on RTX PRO 6000 Blackwell report
Mac mini M4 16 GB8-bit256K23 tok/s4.7 sPDF: Nemotron 3 Nano 4B on Mac mini M4 16 GB report
Mac mini M4 24 GB8-bit256K23 tok/s4.7 sPDF: Nemotron 3 Nano 4B on Mac mini M4 24 GB report
Mac mini M4 32 GB8-bit256K23 tok/s4.7 sPDF: Nemotron 3 Nano 4B on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GB8-bit256K48 tok/s2.3 sPDF: Nemotron 3 Nano 4B on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit256K48 tok/s2.3 sPDF: Nemotron 3 Nano 4B on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit256K48 tok/s2.3 sPDF: Nemotron 3 Nano 4B on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit256K78 tok/s1.2 sPDF: Nemotron 3 Nano 4B on M4 Max 128 GB report
M5 Max 128 GB8-bit256K85 tok/s0.3 sPDF: Nemotron 3 Nano 4B on M5 Max 128 GB report
M3 Ultra 96 GB8-bit256K94 tok/s0.7 sPDF: Nemotron 3 Nano 4B on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit256K94 tok/s0.7 sPDF: Nemotron 3 Nano 4B on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit256K94 tok/s0.7 sPDF: Nemotron 3 Nano 4B on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit256K120 tok/s0.2 sPDF: Nemotron 3 Nano 4B on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit256K37 tok/s0.7 sPDF: Nemotron 3 Nano 4B on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit256K37 tok/s0.7 sPDF: Nemotron 3 Nano 4B on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit256K51 tok/s0.2 sPDF: Nemotron 3 Nano 4B on RTX Spark report
DGX Spark8-bit256K47 tok/s0.2 sPDF: Nemotron 3 Nano 4B on DGX Spark report
Dual DGX Spark8-bit256K47 tok/s0.1 sPDF: Nemotron 3 Nano 4B on Dual DGX Spark report
DGX Station GB3008-bit256K491 tok/s0.1 sPDF: Nemotron 3 Nano 4B on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit256K230 tok/s0.1 sPDF: Nemotron 3 Nano 4B on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit256K230 tok/s0.1 sPDF: Nemotron 3 Nano 4B on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit256K230 tok/s0.1 sPDF: Nemotron 3 Nano 4B on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit256K252 tok/s0.1 sPDF: Nemotron 3 Nano 4B on A100 80 GB SXM report
H100 SXM8-bit256K344 tok/s0.1 sPDF: Nemotron 3 Nano 4B on H100 SXM report
4× H200 NVL8-bit256K415 tok/s0.1 sPDF: Nemotron 3 Nano 4B on 4× H200 NVL report
8× H200 NVL8-bit256K415 tok/s0.1 sPDF: Nemotron 3 Nano 4B 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-nano:4b
llama.cpp
llama-server -hf unsloth/NVIDIA-Nemotron-3-Nano-4B-GGUF
vLLM
vllm serve nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16
MLX (Mac)
mlx_lm.chat --model mlx-community/NVIDIA-Nemotron-3-Nano-4B-4bit

Questions

How much VRAM does Nemotron 3 Nano 4B need?

Nemotron 3 Nano 4B needs 2.2 GB for its weights at 4-bit, 3.2 GB with an 8K context and 3.6 GB with 32K. At 8-bit, 4.9 GB.

What is the minimum VRAM for Nemotron 3 Nano 4B?

About 2.8 GB at 3-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 RTX 4060.

How big is Nemotron 3 Nano 4B?

4B parameters: 7.4 GB of weights at 16-bit, 2.2 GB at 4-bit.

Can Nemotron 3 Nano 4B run on an RTX 4090?

Yes. At 8-bit it holds 256K of context and writes about 148 tokens a second.

More models