Google

Gemma 2

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.

2B

8 GB card
Parameters
2,6B
Q4_K_M file
1,6 Go
Recommended VRAM
≥ 3,1 Go
Max context
8 192

Not measured yet

9B

8 GB card
Parameters
9,2B
Q4_K_M file
5,4 Go
Recommended VRAM
≥ 6,9 Go
Max context
8 192

Not measured yet

27B

24 GB card
Parameters
27,2B
Q4_K_M file
15,5 Go
Recommended VRAM
≥ 17,0 Go
Max context
8 192

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).