The Serving Stack
Start hereCollectionsNewsletter
Sign in

blackwell

1 post

QUANTIZATION · 012.7×DECODE VS BF1616 FP4 VALUES PER BLOCK, ONE SCALE EACH
quantizationnvfp4fp8

NVFP4 on Blackwell: Serving a 4-Bit Model End to End

What NVFP4 stores, how vLLM serves it on Blackwell, and what it changes in memory, speed, and quality.

by The Serving Stack

Inference Notes

Once a week: what changed in serving AI models, and what it means if you build, run or pay for them, in about five minutes.

The Serving Stack

Inference Engineering, In Depth

Navigate

HomeStart hereCollectionsNewsletterAll articles

Collections

Inference EnginesHow an LLM Request Becomes TokensLLMs, ExplainedThe GPU for Inference EngineersQuantization in PracticeServing in Production

Links

RSS Feed

Legal

Privacy PolicyTerms of ServiceCookie Policy
© 2026 The Serving Stack. All rights reserved.