Mistral on NVIDIA Blackwell

Can I run Mixtral 8x7B on an RTX 5070?

Not at Q4_K_M — it needs 28.2 GB against 10.5 GB available. You would need 3 of these cards.

Won't fit

28.2 GB of 10.5 GB · 269%
029 GB
Weights 26.4 GB
KV cache 1.0 GB
Runtime overhead 0.8 GB
Over the limit 17.7 GB

Short by 17.7 GB. You can run it with 10 of 32 layers on the RTX 5070 and the rest in system RAM, at roughly 7.46 tok/s — usable for batch work, painful for chat. A smaller quantisation or a shorter context is usually the better trade.

Generation7.46tok/s
Prompt processing1009tok/s
Max context0tokens
KV per 1K tokens0GB

Every quantisation of Mixtral 8x7B on a RTX 5070

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 1.87
INT8 / W8A8 8.50 46.2 GB 48.0 GB Won't fit 3.81
Q8_0 (GGUF) 8.50 46.2 GB 48.0 GB Won't fit 3.81
FP8 (E4M3) 8.00 43.5 GB 45.3 GB Won't fit 4.04
Q6_K 6.56 35.7 GB 37.5 GB Won't fit 5.04
Q5_K_M 5.67 30.8 GB 32.6 GB Won't fit 5.99
Q5_K_S 5.52 30.0 GB 31.8 GB Won't fit 6.37
Q4_K_M 4.85 26.4 GB 28.2 GB Won't fit 7.46
Q4_K_S 4.58 24.9 GB 26.7 GB Won't fit 8.18
Q4_0 4.55 24.8 GB 26.6 GB Won't fit 8.23
AWQ 4-bit 4.25 23.5 GB 25.3 GB Won't fit 8.65
GPTQ 4-bit 4.25 23.5 GB 25.3 GB Won't fit 8.65
MXFP4 4.25 23.5 GB 25.3 GB Won't fit 8.65
IQ4_XS 4.25 23.1 GB 25.0 GB Won't fit 8.76
Q3_K_M 3.91 21.3 GB 23.1 GB Won't fit 10.3
IQ3_M 3.70 20.2 GB 22.0 GB Won't fit 10.8
IQ3_XXS 3.06 16.7 GB 18.5 GB Won't fit 14.7
Q2_K 2.63 14.4 GB 16.2 GB Won't fit 19.7
IQ2_XXS 2.06 11.3 GB 13.1 GB Won't fit 34.2
IQ1_M 1.75 9.6 GB 11.4 GB Won't fit 57.9

Also worth checking