Mixtral 8x7B Instruct

Mixtral 8x7B Instruct loads in 30 GB at Q4_K_M and is comfortable from 64 GB of memory, Macs included from the MacBook Pro M5 Pro 48GB. Below: the full memory math, the cheapest card that runs it well, and the fastest machine we track.

PARAMETERS
46.7B (13B active)
FORMAT
Q4_K_M
MIN MEMORY
48 GB
BEST FOR
Coding, Quality

Memory math

Weights (Q4_K_M)
30 GB
+ KV cache (16k)
~5.0 GB
Total at 16k
~35.0 GB
Comfortable from
64 GB

KV = fp16 estimate (q8_0 cache roughly halves it). "Comfortable" = weights + KV within the engine's tiered budget (~70-85% of memory).

Hardware snapshot

Cheapest GPU
~12 tok/s est. — ~$3,847 used (as of 2026-07-31)
Fastest
~56 tok/s est.
Runs on a Mac?
~16 tok/s est.

More Mistral models

Mistral 7B InstructMistral Nemo 12BMistral Small 22BMistral Small 3.1Mistral Small 3.1 (Q8)Mistral Medium 3.5

Frequently asked questions

How much memory does Mixtral 8x7B Instruct need?

30 GB for the Q4_K_M weights, plus ~5.0 GB of KV-cache at 16k context — about 35.0 GB total. Comfortable from 64 GB of VRAM or unified memory.

Does Mixtral 8x7B Instruct run on a Mac?

Yes — from the MacBook Pro M5 Pro 48GB (~16 tok/s est.). Unified memory means the RAM budget is the only limit.

What is the cheapest GPU for Mixtral 8x7B Instruct?

The AMD Ryzen AI Max+ 395 (Strix Halo) is the cheapest tracked card that runs Mixtral 8x7B Instruct comfortably — ~12 tok/s est. at ~$3,847 used (as of 2026-07-31).

Cite this page

ModelFit: Mixtral 8x7B Instruct — specs, memory math and hardware verdicts.
https://modelfit.io/models/mixtral-8x7b/ (dataset updated 2026-09-03, CC BY 4.0).