Skip to content
Infra

All guides

Example — not a live product. This page shows the layout for a planned guide; the title, price and contents are placeholders.

GPU sizing for inference

Working title. How to estimate the GPU memory and count you need before you buy or rent.

Example entry for layout only. The idea: a worked approach to sizing GPUs for model serving, from weight memory to KV cache to concurrency.

PricePrice TBA

Placeholder price. Not set yet.

Checkout isn't open. This button doesn't take payment.

Get launch updates instead

Planned topics

Draft outline. It may change before anything is published.

  • Weight memory by parameter count and precision
  • KV cache growth with context length and batch size
  • When to split a model across GPUs
  • Renting versus buying

Other guides in the works