AI Gateway production index
By Steven Van ·
Anthropic leads AI Gateway spend while Google leads token volume, per a report on seven months of production traffic from 200,000+ teams.
Vercel published a production index breaking down how developers actually use AI models through the Vercel AI Gateway, drawing on seven months of production traffic and usage from over 200,000 unique teams.
Spend and volume rankings disagree because they measure different workloads, even for the same customer: premium reasoning calls tend to land on Claude Opus, while cheap, high-volume calls go to Gemini Flash.
- By spend in April 2026, Anthropic took 61%, Google 21%, and OpenAI 12%. OpenAI's spend share tripled from March to April after the GPT-5.4/5.5 releases, and Google's climbed from 8% to 21% over the same period on the strength of Gemini Flash.
- By token volume the picture flips: Google routed 38% of April traffic, Anthropic 26%, OpenAI 13%, and xAI 10%, with smaller labs splitting the rest.
- High-volume workloads route across more than 30 distinct models on average, and open source models are gaining traction with no loyalty to specific labs.
- Agentic workloads now carry 59% of all token volume, up 2x over six months.
The full AI Gateway production index covers tens of trillions of tokens across hundreds of models.
Vercel
The platform for frontend developers — deploy, preview, and scale web apps and AI agents with zero config.
View Vercel →