Skip to content

NVIDIA Nemotron Labs Dory1.0 4B BF16 VRAM requirements

How much VRAM NVIDIA Nemotron Labs Dory1.0 4B BF16 needs
4-bit weights: 2.2 GBWith 8K context: 3.4 GBWith 32K context: 4.3 GB8-bit, 8K context: 5.1 GB
Parameters
4B
Longest context
256K
Released
Oct 2026
License
Openmdw 1.1
Strongest at
Not rated

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

How much VRAM NVIDIA Nemotron Labs Dory1.0 4B BF16 needs

Memory NVIDIA Nemotron Labs Dory1.0 4B BF16 needs at each precision and context length
PrecisionWeightsWith 8KWith 32KWith 128K
16-bit7.4 GB8.6 GB9.4 GB12.8 GB
8-bit3.9 GB5.1 GB5.9 GB9.3 GB
6-bit3.0 GB4.2 GB5.0 GB8.4 GB
5-bit2.6 GB3.8 GB4.6 GB8.0 GB
4-bit2.2 GB3.4 GB4.3 GB7.6 GB
3-bit1.8 GB3.0 GB3.8 GB7.2 GB
2-bit1.4 GB2.6 GB3.4 GB6.8 GB
1.6-bit782 MB1.9 GB2.8 GB6.1 GB

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

GPUs and Macs that run NVIDIA Nemotron Labs Dory1.0 4B BF16

The best way NVIDIA Nemotron Labs Dory1.0 4B BF16 runs on common GPUs, Macs and AI boxes
MachineBest fitContextSpeedFirst replyReport
RTX 40608-bit72K45 tok/s0.5 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 4060 report
RTX 50608-bit72K69 tok/s0.4 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 5060 report
RTX 30708-bit72K69 tok/s0.4 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 3070 report
RTX 3080 10 GB8-bit128K106 tok/s0.3 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 3080 10 GB report
RTX 3060 12 GB8-bit184K57 tok/s0.6 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 3060 12 GB report
Arc B5808-bit184K47 tok/s0.4 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Arc B580 report
RTX 40708-bit184K76 tok/s0.3 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 4070 report
RTX 4070 Super8-bit184K76 tok/s0.2 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 4070 Super report
RTX 50708-bit184K96 tok/s0.3 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 5070 report
RTX 4060 Ti 16 GB8-bit256K47 tok/s0.4 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 4060 Ti 16 GB report
RTX 5060 Ti 16 GB8-bit256K69 tok/s0.3 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 5060 Ti 16 GB report
Arc A770 16 GB8-bit256K55 tok/s0.4 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Arc A770 16 GB report
Radeon RX 9070 XT8-bit256K73 tok/s0.2 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Radeon RX 9070 XT report
RTX 4070 Ti Super8-bit256K96 tok/s0.2 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 4070 Ti Super report
RTX 4080 Super8-bit256K103 tok/s0.2 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 4080 Super report
RTX 5070 Ti8-bit256K120 tok/s0.2 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 5070 Ti report
RTX 50808-bit256K126 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 5080 report
Radeon RX 7900 XTX8-bit256K97 tok/s0.4 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Radeon RX 7900 XTX report
RTX 30908-bit256K124 tok/s0.2 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 3090 report
RTX 40908-bit256K131 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 4090 report
Radeon AI PRO R97008-bit256K73 tok/s0.2 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Radeon AI PRO R9700 report
RTX 50908-bit256K191 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 5090 report
2× RTX 30908-bit256K124 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on 2× RTX 3090 report
2× RTX 40908-bit256K131 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on 2× RTX 4090 report
RTX PRO 5000 Blackwell 48 GB8-bit256K159 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX PRO 5000 Blackwell 48 GB report
RTX 6000 Ada8-bit256K126 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX 6000 Ada report
L40S8-bit256K117 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on L40S report
2× RTX 50908-bit256K191 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on 2× RTX 5090 report
4× RTX 30908-bit256K124 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on 4× RTX 3090 report
RTX PRO 6000 Blackwell8-bit256K191 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX PRO 6000 Blackwell report
Mac mini M4 16 GB8-bit152K22 tok/s5.0 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Mac mini M4 16 GB report
Mac mini M4 24 GB8-bit256K22 tok/s5.0 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Mac mini M4 24 GB report
Mac mini M4 32 GB8-bit256K22 tok/s5.0 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Mac mini M4 32 GB report
Mac mini M4 Pro 24 GB8-bit256K43 tok/s2.5 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Mac mini M4 Pro 24 GB report
Mac mini M4 Pro 48 GB8-bit256K43 tok/s2.5 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Mac mini M4 Pro 48 GB report
Mac mini M4 Pro 64 GB8-bit256K43 tok/s2.5 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Mac mini M4 Pro 64 GB report
M4 Max 128 GB8-bit256K66 tok/s1.3 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on M4 Max 128 GB report
M5 Max 128 GB8-bit256K71 tok/s0.4 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on M5 Max 128 GB report
M3 Ultra 96 GB8-bit256K77 tok/s0.7 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on M3 Ultra 96 GB report
M3 Ultra 256 GB8-bit256K77 tok/s0.7 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on M3 Ultra 256 GB report
M3 Ultra 512 GB8-bit256K77 tok/s0.7 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on M3 Ultra 512 GB report
M5 Ultra 512 GB8-bit256K93 tok/s0.2 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on M5 Ultra 512 GB report
Ryzen AI Max+ 395 64 GB8-bit256K35 tok/s0.7 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Ryzen AI Max+ 395 64 GB report
Ryzen AI Max+ 395 128 GB8-bit256K35 tok/s0.7 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Ryzen AI Max+ 395 128 GB report
RTX Spark8-bit256K49 tok/s0.3 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on RTX Spark report
DGX Spark8-bit256K45 tok/s0.3 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on DGX Spark report
Dual DGX Spark8-bit256K45 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on Dual DGX Spark report
DGX Station GB3008-bit256K340 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on DGX Station GB300 report
2× RTX PRO 6000 Blackwell8-bit256K191 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on 2× RTX PRO 6000 Blackwell report
4× RTX PRO 6000 Blackwell8-bit256K191 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on 4× RTX PRO 6000 Blackwell report
8× RTX PRO 6000 Blackwell8-bit256K191 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on 8× RTX PRO 6000 Blackwell report
A100 80 GB SXM8-bit256K205 tok/s0.2 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on A100 80 GB SXM report
H100 SXM8-bit256K263 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on H100 SXM report
4× H200 NVL8-bit256K302 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B BF16 on 4× H200 NVL report
8× H200 NVL8-bit256K302 tok/s0.1 sPDF: NVIDIA Nemotron Labs Dory1.0 4B 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 cloudnathan5/Dory1.0-4B-GGUF
vLLM
vllm serve nvidia/NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16

Questions

How much VRAM does NVIDIA Nemotron Labs Dory1.0 4B BF16 need?

NVIDIA Nemotron Labs Dory1.0 4B BF16 needs 2.2 GB for its weights at 4-bit, 3.4 GB with an 8K context and 4.3 GB with 32K. At 8-bit, 5.1 GB.

What is the minimum VRAM for NVIDIA Nemotron Labs Dory1.0 4B BF16?

About 3.0 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 NVIDIA Nemotron Labs Dory1.0 4B BF16?

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

Can NVIDIA Nemotron Labs Dory1.0 4B BF16 run on an RTX 4090?

Yes. At 8-bit it holds 256K of context and writes about 131 tokens a second.

More models