LLM / MODEL FILE
Mixtral 8x7B
Mistral model for mixture-of-experts inference. Memory figures are estimates, not measured benchmarks.
Parameters46.7B
Q4 estimate34.1 GB
WorkloadLLM
Best proofTest exact runtime
QUANTIZATION
Memory scenarios
MEMORY FILE
Q4_K_M
common balance of size and quality. Estimated requirement: 34.1 GB.
Inspect →MEMORY FILEQ5_K_M
more quality with a larger footprint. Estimated requirement: 41.4 GB.
Inspect →MEMORY FILEQ8_0
near-full quality with high memory use. Estimated requirement: 59.5 GB.
Inspect →MEMORY FILEFP16
full half-precision weights. Estimated requirement: 117.5 GB.
Inspect →CLOSEST FITS
Hardware near the Q4 line
TIGHT
RTX 5090 32GB
fast 32 GB consumer flagship
32 GB VRAMInspect →OFFLOADRTX 3090 24GB
used-market 24 GB local AI favorite
24 GB VRAMInspect →OFFLOADRTX 4090 24GB
strong 24 GB inference and image generation
24 GB VRAMInspect →OFFLOADRadeon RX 7900 XTX 24GB
24 GB value with a more selective software path
24 GB VRAMInspect →COMFORTABLEMac mini M4 Pro 64GB
compact unified-memory system
64 GB UnifiedInspect →COMFORTABLERTX A6000 48GB
workstation capacity for larger quantized models
48 GB VRAMInspect →Runtime, drivers, memory bandwidth, context, batch size, offloading, and model architecture all matter.