Observed arrival · 2026-11-14
RAMosaurus’s local-AI hardware picker
A set of tools for matching local AI models with computers that can run them, then comparing speed, memory and price.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- People choosing hardware for local AI models
- Worth noticing
- Speed estimates assume one user at roughly 8k context and use 60–80% of usable memory bandwidth, depending on platform.
Field notes
The page lists 45 rigs and 126 local models, with country-specific shop and price selectors; it also notes that the matrix and graph become lists on phones. Its speed method estimates one-user throughput at about 8k context from active model weights and usable memory bandwidth, then adds a fixed per-token cost. Fit estimates account for context and buffers, and developer-reported benchmark scores are distinguished from the named external sources.
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
One card from the complete issue