Massive inference run on Fable cost $1,050 per second, dev reveals
Sauers_ · x · 2026-09-17
Developer Sauers shared a jaw-dropping bill: a massive inference run on Fable was costing $1,050 per second. His self-deprecating reaction aside, the number offers a rare glimpse into just how expensive large-scale inference runs can get.
Related event: Developer Shocked as Large-Scale Inference Run Burns $1,050 Per Second(2 posts)→
More from Fun
- Senior engineers are mad about AI coding because their mastery became a one-minute prompt — FrankFelixAI · 2026-09-17
- Meme Sums Up an AI Gadget: 'Infinite Slop Machine, Portable Casino, 24/7 Surveillance Device' — MarvinTBaumann · 2026-09-17
- "If a Claude-powered autonomous hunter-killer robot dog gets me, it was nice knowing you all" — kevinnbass · 2026-09-17
- A 187M-param zero-shot classifier plays Doom 10x faster with a single forward pass — max_paperclips · 2026-09-17
- Garry Tan shows off his 'personal AGI' with full memory across any harness — garrytan · 2026-09-17
- Chinese drone firm demos cleaning drone by accidentally giving a dog an aerial shower — Polymarket · 2026-09-17