Skip to content

Nemotron Labs Diffusion 3B VRAM requirements

How much VRAM Nemotron Labs Diffusion 3B needs
4-bit weights: 2.2 GBWith 8K context: 3.8 GBWith 32K context: 6.2 GB8-bit, 8K context: 5.4 GB
Parameters
3.8B
Longest context
256K
Released
Mar 2026
License
Nvidia Nemotron Open Model License
Strongest at
Not rated

Its licence (Nvidia Nemotron Open Model License) is not a standard open-source one: check the terms before commercial use.

How much VRAM Nemotron Labs Diffusion 3B needs

Memory Nemotron Labs Diffusion 3B needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit7.1 GB8.7 GB11.2 GB20.9 GB
8-bit3.8 GB5.4 GB7.8 GB17.6 GB
6-bit2.9 GB4.5 GB7.0 GB16.7 GB
5-bit2.5 GB4.1 GB6.6 GB16.3 GB
4-bit2.2 GB3.8 GB6.2 GB16.0 GB
3-bit1.7 GB3.4 GB5.8 GB15.5 GB
2-bit1.3 GB2.9 GB5.4 GB15.1 GB
1.6-bit754 MB2.3 GB4.8 GB14.5 GB

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

GPUs and Macs that run Nemotron Labs Diffusion 3B

The best way Nemotron Labs Diffusion 3B runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 40608-bit28K50 tok/s0.5 sPDF: Nemotron Labs Diffusion 3B on RTX 4060 report
RTX 50608-bit28K81 tok/s0.4 sPDF: Nemotron Labs Diffusion 3B on RTX 5060 report
RTX 30708-bit28K81 tok/s0.3 sPDF: Nemotron Labs Diffusion 3B on RTX 3070 report
RTX 3080 10 GB8-bit48K131 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on RTX 3080 10 GB report
RTX 3060 12 GB8-bit64K66 tok/s0.5 sPDF: Nemotron Labs Diffusion 3B on RTX 3060 12 GB report
Arc B5808-bit64K57 tok/s0.4 sPDF: Nemotron Labs Diffusion 3B on Arc B580 report
RTX 40708-bit64K90 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on RTX 4070 report
RTX 4070 Super8-bit64K90 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on RTX 4070 Super report
RTX 50708-bit64K117 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on RTX 5070 report
RTX 4060 Ti 16 GB8-bit104K53 tok/s0.3 sPDF: Nemotron Labs Diffusion 3B on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GB8-bit104K81 tok/s0.3 sPDF: Nemotron Labs Diffusion 3B on RTX 5060 Ti 16 GB report
Arc A770 16 GB8-bit104K69 tok/s0.3 sPDF: Nemotron Labs Diffusion 3B on Arc A770 16 GB report
Radeon RX 9070 XT8-bit104K93 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on Radeon RX 9070 XT report
RTX 4070 Ti Super8-bit104K117 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on RTX 4070 Ti Super report
RTX 4080 Super8-bit104K127 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on RTX 4080 Super report
RTX 5070 Ti8-bit104K152 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on RTX 5070 Ti report
RTX 50808-bit104K161 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on RTX 5080 report
Radeon RX 7900 XTX8-bit176K131 tok/s0.3 sPDF: Nemotron Labs Diffusion 3B on Radeon RX 7900 XTX report
RTX 30908-bit176K158 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on RTX 3090 report
RTX 40908-bit176K168 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on RTX 4090 report
Radeon AI PRO R97008-bit248K93 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on Radeon AI PRO R9700 report
RTX 50908-bit248K271 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on RTX 5090 report
2× RTX 30908-bit256K158 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on 2× RTX 3090 report
2× RTX 40908-bit256K168 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit256K215 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit256K161 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on RTX 6000 Ada report
L40S8-bit256K147 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on L40S report
2× RTX 50908-bit256K271 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on 2× RTX 5090 report
4× RTX 30908-bit256K158 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit256K271 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on RTX PRO 6000 Blackwell report
Mac mini M4 16 GB8-bit56K26 tok/s4.4 sPDF: Nemotron Labs Diffusion 3B on Mac mini M4 16 GB report
Mac mini M4 24 GB8-bit104K26 tok/s4.4 sPDF: Nemotron Labs Diffusion 3B on Mac mini M4 24 GB report
Mac mini M4 32 GB8-bit152K26 tok/s4.4 sPDF: Nemotron Labs Diffusion 3B on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GB8-bit104K54 tok/s2.2 sPDF: Nemotron Labs Diffusion 3B on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit256K54 tok/s2.2 sPDF: Nemotron Labs Diffusion 3B on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit256K54 tok/s2.2 sPDF: Nemotron Labs Diffusion 3B on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit256K91 tok/s1.1 sPDF: Nemotron Labs Diffusion 3B on M4 Max 128 GB report
M5 Max 128 GB8-bit256K100 tok/s0.3 sPDF: Nemotron Labs Diffusion 3B on M5 Max 128 GB report
M3 Ultra 96 GB8-bit256K112 tok/s0.6 sPDF: Nemotron Labs Diffusion 3B on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit256K112 tok/s0.6 sPDF: Nemotron Labs Diffusion 3B on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit256K112 tok/s0.6 sPDF: Nemotron Labs Diffusion 3B on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit256K148 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit256K40 tok/s0.6 sPDF: Nemotron Labs Diffusion 3B on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit256K40 tok/s0.6 sPDF: Nemotron Labs Diffusion 3B on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit256K55 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on RTX Spark report
DGX Spark8-bit256K50 tok/s0.2 sPDF: Nemotron Labs Diffusion 3B on DGX Spark report
Dual DGX Spark8-bit256K50 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on Dual DGX Spark report
DGX Station GB3008-bit256K661 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit256K271 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit256K271 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit256K271 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit256K300 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on A100 80 GB SXM report
H100 SXM8-bit256K429 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on H100 SXM report
4× H200 NVL8-bit256K536 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B on 4× H200 NVL report
8× H200 NVL8-bit256K536 tok/s0.1 sPDF: Nemotron Labs Diffusion 3B 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 IntellAgents/Nemotron-Labs-Diffusion-3B-GGUF
vLLM
vllm serve nvidia/Nemotron-Labs-Diffusion-3B
MLX (Mac)
mlx_lm.chat --model mlx-community/Nemotron-Labs-Diffusion-3B-4bit

Questions

How much VRAM does Nemotron Labs Diffusion 3B need?

Nemotron Labs Diffusion 3B needs 2.2 GB for its weights at 4-bit, 3.8 GB with an 8K context and 6.2 GB with 32K. At 8-bit, 5.4 GB.

What is the minimum VRAM for Nemotron Labs Diffusion 3B?

About 3.4 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 Labs Diffusion 3B?

3.8B parameters: 7.1 GB of weights at 16-bit, 2.2 GB at 4-bit.

Can Nemotron Labs Diffusion 3B run on an RTX 4090?

Yes. At 8-bit it holds 176K of context and writes about 168 tokens a second.

More models