Which rig runs which local AI model, per dollar.

36 GPUs and 30 models with cited specs. Speeds are labeled estimates from memory bandwidth until measured results come in.

Budget + model picker

Budget Model

Speed calculator estimate

GPU Count Model

Estimate = 0.76 × bandwidth ÷ GB read per token (Q4 file size, scaled by active/total parameters for MoE). Fit = file size + 1.5 GB headroom ≤ total VRAM. Multi-GPU: VRAM adds, speed uses one card's bandwidth (layer split). Short context only. How this works

Which rig runs which local AI model, per dollar.

Budget + model picker

Speed calculator estimate

Popular questions