CALCULATOR
LLM VRAM calculator
Estimate model weights and runtime reserve by parameter count and quantization.
The interactive matcher supports Q4, Q5, Q8, FP16, dedicated VRAM, and unified-memory assumptions.
LAUNCH MATCHER →Estimate model weights and runtime reserve by parameter count and quantization.
The interactive matcher supports Q4, Q5, Q8, FP16, dedicated VRAM, and unified-memory assumptions.
LAUNCH MATCHER →