QDNASales and integration of LLM inference and training platforms, on-premises or hybrid

Quantisation: memory by format

The weight format decides how much memory is needed, more surely than the model itself.

Short answer. Moving from FP16 to NVFP4 divides the weight footprint by 4. 28 pages detail each model in each format.

How much memory per format?

The figure covers the weights, runtime margin included, excluding the attention cache.

ModelBillion parameters FP16FP8NVFP4Q4
GLM 5.27441,711 GB856 GB428 GB428 GB
Kimi K328006,440 GB3,220 GB1,610 GB1,610 GB
Kimi K2.7 Code10002,300 GB1,150 GB575 GB575 GB
DeepSeek V416003,680 GB1,840 GB920 GB920 GB
Nemotron 3 Ultra5501,265 GB632 GB316 GB316 GB
MiniMax M3428984 GB492 GB246 GB246 GB
Qwen 3.6 27B2762.1 GB31.0 GB15.5 GB15.5 GB

What does quantisation cost?

  • FP16: full precision, the quality reference.
  • FP8: negligible loss on most tasks.
  • NVFP4: Blackwell format, native FP4 compute.
  • Q4: block quantisation, broadly supported.

Method

Footprints are calculated from the parameter count and the format, with a runtime margin. They are not measured on hardware. The attention cache sits on top and depends on context and concurrency. Prices are indicative and not contractual.