Mistral on NVIDIA Hopper

Can I run Mixtral 8x7B on an H100 SXM 80GB?

Yes. Mixtral 8x7B at Q4_K_M uses 28.2 GB of the 74.4 GB available on a H100 SXM 80GB, and runs at about 220 tokens per second. You can push the context to 32K.

Runs comfortably

28.2 GB of 74.4 GB · 38%
074 GB
Weights 26.4 GB
KV cache 1.0 GB
Runtime overhead 0.8 GB

Mixtral 8x7B at Q4_K_M leaves 46.2 GB spare on a H100 SXM 80GB. There is room to raise the context length or move up a quantisation level.

Generation220tok/s
Prompt processing16100tok/s
Max context32Ktokens
KV per 1K tokens0GB

Every quantisation of Mixtral 8x7B on a H100 SXM 80GB

Highlighted row is the highest quality that still fits at 8K context.

QuantisationbpwWeights TotalVerdictMax ctxtok/s
FP16 / BF16 16.00 87.0 GB 88.8 GB Won't fit — 8.82
INT8 / W8A8 8.50 46.2 GB 48.0 GB Runs comfortably 32K 138
Q8_0 (GGUF) 8.50 46.2 GB 48.0 GB Runs comfortably 32K 138
FP8 (E4M3) 8.00 43.5 GB 45.3 GB Runs comfortably 32K 145
Q6_K 6.56 35.7 GB 37.5 GB Runs comfortably 32K 172
Q5_K_M 5.67 30.8 GB 32.6 GB Runs comfortably 32K 194
Q5_K_S 5.52 30.0 GB 31.8 GB Runs comfortably 32K 199
Q4_K_M 4.85 26.4 GB 28.2 GB Runs comfortably 32K 220
Q4_K_S 4.58 24.9 GB 26.7 GB Runs comfortably 32K 230
Q4_0 4.55 24.8 GB 26.6 GB Runs comfortably 32K 232
AWQ 4-bit 4.25 23.5 GB 25.3 GB Runs comfortably 32K 242
GPTQ 4-bit 4.25 23.5 GB 25.3 GB Runs comfortably 32K 242
MXFP4 4.25 23.5 GB 25.3 GB Runs comfortably 32K 242
IQ4_XS 4.25 23.1 GB 25.0 GB Runs comfortably 32K 244
Q3_K_M 3.91 21.3 GB 23.1 GB Runs comfortably 32K 260
IQ3_M 3.70 20.2 GB 22.0 GB Runs comfortably 32K 271
IQ3_XXS 3.06 16.7 GB 18.5 GB Runs comfortably 32K 311
Q2_K 2.63 14.4 GB 16.2 GB Runs comfortably 32K 345
IQ2_XXS 2.06 11.3 GB 13.1 GB Runs comfortably 32K 404
IQ1_M 1.75 9.6 GB 11.4 GB Runs comfortably 32K 445

Also worth checking

Mixtral 8x7B on every GPU
Full hardware tableEverything that runs on a H100 SXM 80GB
Full model tableWhy the KV cache is the part that bites
Guide