DeepSeek

Coder V2 Lite

codeMoE

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.

16B-A2.4B

12 GB card
Parameters
15,7B · 2,4B active
Q4_K_M file
9,7 Go
Recommended VRAM
≥ 11,2 Go
Max context
163 840

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).