Mapping LLM Size Tiers to Hardware Specs
DeepOrangeSky · reddit · 2026-07-09
The post asks what hardware constraints dictate common local LLM size tiers (like 30B, 70B, 120B, 230B). Is it based on professional GPUs, consumer GPU VRAM, or a mixed consideration post-quantization?
The author specifically wants to understand why these common sizes have formed fixed "tiers" and what server or desktop deployment conditions these tiers correspond to.
More from Infra
- AI datacenters hit diseconomies of scale as inference shifts demand smaller — abhiadesai · 2026-07-22
- Nothing phone mockup turns a film joke into a modular design meme — ZeYanjie · 2026-07-22
- Actual Computer says its inference stack is tuned for Nvidia’s consumer Blackwell lineup — markjeffrey · 2026-07-22
- Ben Bajarin says CPU demand is still being badly underestimated — BenBajarin · 2026-07-22
- An energy model says the U.S. could run short of natural gas starting in 2028 — churchkey · 2026-07-22
- Devin adds e2b sandboxes for remote agent execution — badphilosopher · 2026-07-22