DeepSeek overtakes Google on volume, cost per token falls 13.6%
By Steven Van ·
AI Gateway Production Index — August 2026 Every month, AI Gateway routes tens of trillions of tokens between production applications and AI labs.
Vercel's monthly Vercel AI Gateway Production Index, covering usage through July 2026, found the average price paid per token falling 13.6% even as spend across the gateway grew 37%. The full breakdown is in the original report.
- DeepSeek is now the second-largest lab by token volume on the gateway, running more than twice Google's volume; its cheapest model, V4 Flash, processed more tokens on its own than all of Google's models combined.
- Anthropic collected 65.1% of gateway spending on 30% of token volume, at 4.4 times the average price per token of every other lab, up from 3.4 times in June.
- Open-weight models' share of gateway spend more than doubled in July to 8.6%, mostly driven by Moonshot's Kimi K3 (released July 16) and Z.ai's GLM 5.2, both built for long-horizon agent work.
- Google's Nano Banana took the lead in image volume from OpenAI's GPT Image, and ByteDance's Seedance led video in both volume and dollars.
Kimi K3's daily volume tripled between launch week and the end of July, and by month's end its requests were as heavy as Claude Opus 4.8's, ranking eighth on the gateway by token volume on the last full day of the month.
Vercel
The platform for frontend developers — deploy, preview, and scale web apps and AI agents with zero config.
View Vercel →