---
format: "aidr-story-markdown/v1"
id: "e70e960cf95d16952c679373a51073d057588c98b5a5fb20b1a61ab1fc977c89"
canonical_url: "https://aidr.today/e70e960c?lang=en"
title: "Show HN: Sunk Cost – How long until a local LLM rig pays for itself?"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-09-15T01:37:43.000Z"
category: "Infra"
topics: ["llm","open-source","chips","inference","infra"]
source_urls: ["https://sunkcost.ai/","https://news.ycombinator.com/item?id=49706656"]
summary: "Which open models fit your machine, how fast they run, and how many tokens it takes to pay back the hardware against an API. Honest enough to say never."
---

# Show HN: Sunk Cost – How long until a local LLM rig pays for itself?

> [Open the canonical story](<https://aidr.today/e70e960c?lang=en>)

**Published:** 2026-09-15T01:37:43.000Z
**Category:** Infra
**Topics:** llm, open\-source, chips, inference, infra

## Summary

Which open models fit your machine, how fast they run, and how many tokens it takes to pay back the hardware against an API\. Honest enough to say never\.

## Sources

- [Story source](<https://sunkcost.ai/>)
- [Discussion](<https://news.ycombinator.com/item?id=49706656>)