Local LLM model fit

Can my GPU run Mixtral 8x7B?

Mixtral 8x7B is a 46.7B Mixtral model. This page estimates Q4 VRAM fit, Ollama command, context planning, and fallback choices for common local AI GPUs.

Check Mixtral 8x7B in the calculator

Q4 runtime estimate28 GB
Ollama commandollama run mixtral:8x7b
Recommended GPU32GB+ VRAM or CPU/RAM offload

Best use

MoE local reasoning/chat when enough VRAM is available. Weakness: Not practical for 24GB single-GPU setups without offload.

GPU fit table

HardwareExamplesClean capacityQ4 needStatusCalculator
6 GB VRAM entry GPUGTX 1660, RTX 2060 6GB4.5 GB clean VRAM28 GBToo largeOpen calculator
8 GB VRAM mainstream GPURTX 3060 Ti, RTX 4060, RTX 30706.5 GB clean VRAM28 GBRAM offloadOpen calculator
10 GB VRAM older high-end GPURTX 3080 10GB8.5 GB clean VRAM28 GBRAM offloadOpen calculator
12 GB VRAM local agent GPURTX 3060 12GB, RTX 4070, RTX 507010.5 GB clean VRAM28 GBRAM offloadOpen calculator
16 GB VRAM creator GPURTX 4060 Ti 16GB, RTX 4080, RTX 5070 Ti, RTX 508014.5 GB clean VRAM28 GBRAM offloadOpen calculator
24 GB VRAM homelab workstationRTX 3090, RTX 409022.5 GB clean VRAM28 GBRAM offloadOpen calculator
32 GB VRAM Blackwell workstationRTX 509030.5 GB clean VRAM28 GBRuns locallyOpen calculator
48 GB VRAM workstationRTX A6000, L40S 48GB46.5 GB clean VRAM28 GBRuns locallyOpen calculator
Apple Silicon 32 GB unified memoryM2 Max 32GB, M3 Max 36GB26 GB unified28 GBToo largeOpen calculator
Apple Silicon 256 GB unified memoryMac Studio M3 Ultra 256GB, Mac Studio M4 Ultra 256GB250 GB unified28 GBRuns locallyOpen calculator

Quantization memory estimate on a 12GB GPU preset

QuantizationEstimated memoryUse case
Q4 / 4-bit28 GBDefault local inference balance
Q5 / 5-bit35 GBBetter quality, more VRAM
Q8 / 8-bit56 GBHigh quality, much more VRAM
FP16 / 16-bit112 GBMostly workstation/server use

Data sources and confidence

This is a practical planning estimate, not a benchmark. Real memory use changes with backend, context length, KV cache, quantization file, drivers, and offloading settings.

Verified

2026-05-19

Confidence

high