Home / Will it run? / Codestral 22B v0.1
Mistral AI · Codestral
Will Codestral 22B v0.1 run on my GPU?
22.2B parameters, 32,768-token context, Mistral AI Non-Production License. At Q4_K_M the weights come to about 12.4 GiB; add 1.5 GiB for the runtime and you need roughly 13.9 GiB of VRAM before context.
Weights by quant
GiBQ4_K_M · the everyday quant12.4 GiB
Q8_0 · near-lossless21.9 GiB
FP16 · full weights41.4 GiB
Runtime reserve1.5 GiB
Card by card
11 of 17 fit at Q4Headroom is what remains for the KV cache after weights and the 1.5 GiB reserve. How far it stretches depends on the architecture: grouped-query models are frugal, older dense models are not. If the table says tight, plan on a shorter context or a lower quant. Parameter count from the model's own card.