Liquid AI · Hugging Face

LFM2.5

small machinesMoE

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.

1.2B

8 GB card
Parameters
1,2B
Q4_K_M file
0,7 Go
Recommended VRAM
≥ 2,2 Go
Max context
128 000

Not measured yet

2.6B

8 GB card
Parameters
2,7B
Q4_K_M file
1,6 Go
Recommended VRAM
≥ 3,1 Go
Max context
131 072

Not measured yet

8B-A1B

8 GB card
Parameters
8,5B · 1B active
Q4_K_XL file
5,0 Go
Recommended VRAM
≥ 6,5 Go
Max context
128 000

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).