Observed arrival · 2026-10-13
Ankit Dongare’s LLM GPU cost bench
An AI engineer’s site pairs GPU price comparisons and inference writing with a calculator estimating how many GPUs a model needs and what they cost per hour.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- People choosing GPUs to run language models
- Worth noticing
- The calculator rounds GPU counts to 1, 2, 4, or 8 for tensor-parallel splits; listed rates may not be in stock.
Field notes
The calculator exposes model sizes from 8B to 405B parameters and precision choices of FP16, FP8, and INT4. Its GPU-count estimate rounds to powers of two up to eight, while the price board uses the lowest listed on-demand rate per GPU-hour from getdeploying.com; the page cautions that some offers may not be available. Several further benchmark topics are labeled in progress.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
$PaidCommerce or pricing visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
◉HumanPersonal, local, civic, or handmade
One card from the complete issue