Open-weight models surge to 29% of volume, price per token flattens
By Steven Van ·
Anthropic still captured 61% of gateway spend, and a US export-control order cut short Claude Fable 5's June launch.
Vercel's AI Gateway processed 29% more tokens in June 2026 than the month before, with spend up 27%, according to the latest AI Gateway production index. The average price per token held flat, after rising almost 20% in May.
Open-weight models drove the volume growth: they ran 29% of gateway tokens in June, up from 11% in April, while taking under 4% of spend.
- DeepSeek reached 22.6% of token volume, putting it in third place and within two points of Google, and GLM 5.2 broke into the gateway's top models by volume two weeks after release.
- Anthropic took 61% of gateway spend on 32% of tokens, capturing 72% or more of spend in every high-stakes use case.
- Among image and video models, OpenAI's GPT Image generated 53% of images against Google's Nano Banana at 39%, Chinese labs took two-thirds of video spend, and xAI's Grok Imagine generated 42% of videos, more than any other model.
- Anthropic's Claude Fable 5, released June 9, reached 22% of Opus 4.8's request volume within four days, before a US export-control directive suspended access for the rest of the month.
The flat price per token hides two offsetting moves: open-weight tokens now cost about a tenth of the gateway average, but closed-weight frontier prices rose about 12% per token over the same month, cancelling out the effect of the cheaper volume.
Vercel
The platform for frontend developers — deploy, preview, and scale web apps and AI agents with zero config.
View Vercel →