Sunk Cost: A calculator for when a local LLM rig pays for itself vs renting tokens
rlindsey123 · hn · 2026-09-15
Developer rlindsey123 built Sunk Cost, a tool to test the claim that "just buy a Mac and run models locally — it pays for itself." Given a machine, a model, and daily token usage, it calculates how long the hardware takes to break even against renting the same model by the token. The author notes it's purely a cost-saving analysis and excludes other reasons to own hardware, and is seeking feedback on making it more useful.
More from Infra
- Video of New Albany, Ohio data centers stuns: scale is 'staggering' — natesiggard · 2026-09-15
- "This will happen to inference": Suhail bets LLM inference costs will fall 1000x like lighting did — npew · 2026-09-15
- HTTP 200 is a weak success metric for AI endpoints, engineer warns — gethackteam · 2026-09-15
- US's #2 law firm Latham & Watkins is building its own in-house Nvidia AI stack — MikeBirdTech · 2026-09-15
- Custom llama.cpp fork pushes 27B models to 90+ TPS on an RTX 3090 — Brief-Tap-6616 · 2026-09-15
- GLM 5.3 Flash EXL3 adds TP=3 mode: 28% faster decode across three DGX Sparks — HankYeomans · 2026-09-15