Home / Will it run? / Mixtral 8x7B Instruct v0.1
Mistral AI · Mixtral
Will Mixtral 8x7B Instruct v0.1 run on my GPU?
46.7B parameters (mixture of experts, 12.9B active per token, but every expert must sit in memory), 32,768-token context, Apache-2.0. At Q4_K_M the weights come to about 26.1 GiB; add 1.5 GiB for the runtime and you need roughly 27.6 GiB of VRAM before context.
Weights by quant
GiBQ4_K_M · the everyday quant26.1 GiB
Q8_0 · near-lossless46.1 GiB
FP16 · full weights87.0 GiB
Runtime reserve1.5 GiB
Card by card
1 of 18 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. The Buy column opens live Amazon listings; those links may earn us a commission.