← All posts

Khoury College of Computer Sciences

CS 4550 / CS 5610 · Web Development · Fall 2026

AI Gateway Index: open-weight models hit 56% of token volume

  • AI
  • Course

Instructor-curated digest, not original reporting. Each post paraphrases a public article and links to the original source.

On September 17, 2026, Amelia Charles and Eric Dodds published the September AI Gateway Production Index, covering data through August 2026. Open-weight models ran 56% of AI Gateway tokens in August, up from 7% in December — the first month they held a majority. Average price per token fell 23.2% in August, the third straight monthly drop; among teams with more than 10M tokens in both months, the median team paid 7.6% less.

On frontier spend, Fable 5 share of gateway spend fell from 13.2% in July to 4.9% in August as Opus 5 rose to 22.5%, while Anthropic kept 64% of all spend. A special report notes GPT-6 Astra launched on Gateway September 3 and within two days took a third of OpenAI spend; over its first twelve days Astra was 7.7% of all gateway spend versus Fable 5.1 at 3.7% (Fable 5.1 launched September 1 at the same price). A September 18 note in the post adds that Jev became the fastest-adopted model in Gateway history, reaching about 13% of paid teams in the first 24 hours.

Those Gateway numbers come from the same Vercel platform you already deploy to in Chapter 1 (and the Node server work in Chapter 5).

Read original: Open-weight models take 56% of token volume, Astra doubles Fable 5.1 spend — Vercel

Related in the book: Chapter 1 · Chapter 5