local intelligence

can it run

Can the Tesla V100 32GB (used) run mixtral 8x7B MoE?

STRETCH

Only with compromises. 32GB is below the 33GB working set — expect a lower quant, offloading, or heavy swap. Possible, not pleasant.

~110–170 tok/s (active-expert estimate)

weights 26GB + KV cache 0.5GB at 4k context vs ~32GB usable → tight. 26.5 of 32 GB usable — keep context modest

computed roofline: 897 GB/s × 0.46–0.76 efficiency window / 3.9GB (Q4_K_M) — bands, never points.

MoE: file is 26GB (that decides fit) but only ~1.05B of experts activate per token (~3.9GB streamed) — band below is the active-expert estimate, optimistic; community MoE rows run several× the naive bandwidth math

explore the full catalog90 machines indexed · live prices · what each one can run

Specs

model file26GB (GGUF Q4_K_M)
minimum memory33GB working set
this machine32GB HBM2, 897 GB/s
params8x7B MoE
licenseapache-2.0
price$645

Run it

model files: hugging face ↗

easy run: ollama pull mixtral:8x7b

runs with: ollama · llama.cpp · koboldcpp · lm studio

Alternatives

Mac mini M5 Pro (64GB) — cheapest machine that runs it ($1669)

ASUS ExpertCenter Pro ET900N G3 (GB300) — this model on the biggest machine

every model the Tesla V100 32GB (used) can run — full list

RTX 3090 (24GB, used) — upgrade path machine

Buy

eBay (used) ↗ · Amazon ↗

2026-09-20 · ← all models on the Tesla V100 32GB (used) · ← full hardware catalog · prices verified at source, may drift