Skip to content

Head-to-head · Mistral

Mixtral 8x7B vs Devstral 2 123B

Mixtral 8x7B needs ~28.9 GB at Q4_K_M; Devstral 2 123B needs ~73.3 GB. That ~44.4 GB gap decides which hardware runs each; they differ on 1 of the 11 devices below.

Spec Mixtral 8x7B Devstral 2 123B
Parameters46.7B MoE123B
Memory at Q4_K_M~28.9 GB~73.3 GB
Context window32k256k
LMArenanot rankednot ranked
LicenseApache-2.0Modified MIT

Which devices run each

A representative spread across the memory range. Tap a verdict for the full breakdown.

Size at each quantization

Quant Mixtral 8x7B Devstral 2 123B
Q2_K 19.6 GB* 51.5 GB*
Q3_K_M 22.8 GB* 60.1 GB*
Q4_K_M 26.49 GB 69.75 GB
Q5_K_M 33.3 GB* 87.6 GB*
Q6_K 38.3 GB* 100.9 GB*
Q8_0 46.22 GB 123.73 GB
FP16 93.4 GB* 250.06 GB

* derived from bits-per-weight; unstarred sizes are measured GGUF files.

Bottom line

The larger Devstral 2 123B needs ~73.3 GB; Mixtral 8x7B runs on lighter hardware at ~28.9 GB with more headroom and faster responses. Check each against your exact device: what runs Mixtral 8x7B · what runs Devstral 2 123B.

More Mistral comparisons

FAQ

What is the difference in memory between Mixtral 8x7B and Devstral 2 123B?

At Q4_K_M, Mixtral 8x7B needs about 28.9 GB and Devstral 2 123B needs about 73.3 GB, a difference of ~44.4 GB.

Should I run Mixtral 8x7B or Devstral 2 123B?

Run Devstral 2 123B if your hardware has the ~73.3 GB it needs and you want maximum quality; run Mixtral 8x7B (~28.9 GB) for lighter hardware, faster responses, and more memory headroom.

Full breakdowns: Mixtral 8x7B · Devstral 2 123B · all models × devices.

Sources

Memory figures are estimates at Q4_K_M, validated 2026-09-07. See methodology.