NVIDIA

Nemotron 3

MoE

Pick a size: the video memory shown is the Q4_K_M file (the protocol's quantisation) plus about 1.5 GB for a 4,000-token context.

Nano 4B

8 GB card
Parameters
4B
Q4_K_M file
2,7 Go
Recommended VRAM
≥ 4,2 Go
Max context
262 144

Not measured yet

Nano 30B-A3B

32 GB card
Parameters
31,6B · 3B active
Q4_K_M file
22,8 Go
Recommended VRAM
≥ 24,3 Go
Max context
262 144

Not measured yet

Super 120B-A12B

> 32 GB
Parameters
123,6B · 12B active
Q4_K_XL file
78,0 Go
Recommended VRAM
≥ 79,5 Go
Max context
262 144

Not measured yet

MoE (“30B-A3B”): all the memory of a 30B, but the speed of a model with 3B active parameters. “> 32 GB”: several cards, or part of the model in system memory (much slower).