Fireworks serves over 40T tokens daily, all from specialized post-trained models
andrew_n_carr · x · 2026-11-04
Nat Lambert shared that inference provider Fireworks AI serves over 40 trillion tokens per day, all from "specialized models" — post-trained models customized for individual customers rather than general-purpose APIs. Andrew Carr expressed surprise and asked about the growth curve and whether that token volume is actually large, hinting at the scale of enterprise custom-model inference demand.
More from Companies & People
- Xudong Sun wins honorable mention for SIGOPS Dennis M. Ritchie Dissertation Award — LingmingZhang · 2026-11-04
- Uber President Jill Hazelbaker joins OpenAI as chief corporate affairs officer — steph_palazzolo · 2026-11-04
- Ex-Meta founder launches Private SuperIntelligence AI concierge fed 20 years of his data — Scobleizer · 2026-11-04
- Elvis Saravia: enterprises should adopt RSI gradually, from small functions to high-stakes processes — omarsar0 · 2026-11-04
- CMU opens tenure-track faculty search in computational biology including AI for biology — jmuiuc · 2026-11-04
- OpenAI employees warned leadership for years about security lapses and espionage risks — GarrisonLovely · 2026-11-04