Mistral on NVIDIA Ada

Can I run Mixtral 8x7B on an RTX 4060 Ti 16GB?

Not at Q4_K_M — it needs 28.2 GB against 14.2 GB available. You would need 2 of these cards.

Won't fit

28.2 GB of 14.2 GB · 198%
029 GB
Weights 26.4 GB
KV cache 1.0 GB
Runtime overhead 0.8 GB
Over the limit 14.0 GB

Short by 14.0 GB. You can run it with 15 of 32 layers on the RTX 4060 Ti 16GB and the rest in system RAM, at roughly 7.50 tok/s — usable for batch work, painful for chat. A smaller quantisation or a shorter context is usually the better trade.

Generation7.50tok/s
Prompt processing716tok/s
Max context0tokens
KV per 1K tokens0GB

Every quantisation of Mixtral 8x7B on a RTX 4060 Ti 16GB

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 1.72
INT8 / W8A8 8.50 46.2 GB 48.0 GB Won't fit 3.54
Q8_0 (GGUF) 8.50 46.2 GB 48.0 GB Won't fit 3.54
FP8 (E4M3) 8.00 43.5 GB 45.3 GB Won't fit 3.86
Q6_K 6.56 35.7 GB 37.5 GB Won't fit 4.95
Q5_K_M 5.67 30.8 GB 32.6 GB Won't fit 5.86
Q5_K_S 5.52 30.0 GB 31.8 GB Won't fit 6.21
Q4_K_M 4.85 26.4 GB 28.2 GB Won't fit 7.50
Q4_K_S 4.58 24.9 GB 26.7 GB Won't fit 7.91
Q4_0 4.55 24.8 GB 26.6 GB Won't fit 8.25
AWQ 4-bit 4.25 23.5 GB 25.3 GB Won't fit 8.67
GPTQ 4-bit 4.25 23.5 GB 25.3 GB Won't fit 8.67
MXFP4 4.25 23.5 GB 25.3 GB Won't fit 8.67
IQ4_XS 4.25 23.1 GB 25.0 GB Won't fit 9.11
Q3_K_M 3.91 21.3 GB 23.1 GB Won't fit 10.2
IQ3_M 3.70 20.2 GB 22.0 GB Won't fit 11.2
IQ3_XXS 3.06 16.7 GB 18.5 GB Won't fit 15.9
Q2_K 2.63 14.4 GB 16.2 GB Won't fit 22.5
IQ2_XXS 2.06 11.3 GB 13.1 GB Fits, but tight 17K 39.9
IQ1_M 1.75 9.6 GB 11.4 GB Runs comfortably 30K 45.6

Also worth checking