QDNAAdvisory and architecture for LLM inference and training platforms, on-premises or hybrid

Sectors: local LLM and compliance

The regulatory framework varies by sector. The question of data transfer does not.

Short answer. A model served on site processes data in place. No data leaves the information system, which removes the question of transfer to a third party.

By sector

What these pages do not say

They describe an architecture and are not legal advice. Compliance is assessed on the actual processing, with the organisation’s data protection officer.

Method and sources

Footprints are calculated: billion parameters × bytes per parameter of the format × 1.15 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. Each detail page carries its sources; the shared conventions are these.

  • Bytes per parameter: FP16 2 (16 bits per parameter, hence 2 bytes); FP8 1 (8 bits per parameter (E4M3), hence 1 byte, block scales not counted); NVFP4 0.5 (4 bits per parameter, hence 0.5 byte; 4-bit published weights weigh 0.54 to 0.56 byte per parameter with their scales (DeepSeek V4 Pro 865 GB for 1,599 billion, Kimi K3 1,561 GB for 2,780 billion)); Q4 0.6 (taken here as llama.cpp Q4_K_M, 0.6 byte per parameter including scales; Q4_K_S weighs 0.56, AWQ and GPTQ 0.55). Source: connaissance/faits.yaml, quantifications family, checked on 2026-08-31.
  • Runtime margin × 1.15: QDNA operating assumption, not measured: 15% above the weights for activations, buffers and fragmentation.
  • “Fits” threshold at 75% of memory: QDNA assumption, not measured, which keeps the remaining quarter for the attention cache and concurrency.