Massive inference run costs $1,050 per second, researcher shares sticker shock
Sauers_ · x · 2026-09-17
Sauers shared that a massive inference run was costing roughly $1,050 per second according to Fable — a striking look at the extreme cost of large-scale frontier inference.
Related event: Developer Shocked as Large-Scale Inference Run Burns $1,050 Per Second(2 posts)→
More from Infra
- Boson AI CEO Alexander Smola on Voice Agents, Avatars, Latency and Emotional Intelligence — TWIML AI Podcast · 2026-09-17
- DDRop attack drops DDR5 writes to break Intel TDX and AMD SEV-SNP confidential VMs — matthew_d_green · 2026-09-17
- PufferLib 5.0 hits 60M steps/sec single-GPU RL training, solves Breakout in under a second — jsuarez · 2026-09-17
- Crusoe signs multiyear cloud deal with Perplexity to expand AI chip rentals — inductionheads · 2026-09-17
- SGLang's Delta Router Replay slashes sync stalls in Kimi K2 agentic RL training — hsu_byron · 2026-09-17
- SemiAnalysis: Rubin NVL72 delivers 7x perf-per-watt over GB300, far above Jensen's 3x claim — inductionheads · 2026-09-17