The Agent RaceJuly 13, 2026via Vercel Blog
Open-weight models surge to 29% of volume, price per token flattens
Why it matters
Enterprise production data shows a dramatic shift in AI model adoption: open-weight models (especially DeepSeek) are capturing nearly a third of token volume at 1/25th the cost of closed models, while frontier labs are doubling down on high-stakes workloads. This signals a fundamental market segmentation where cheap open models handle scale and closed models command premium pricing for mission-critical tasks.
Key signals
- Open-weight models: 29% of token volume, under 4% of spend (June 2026)
- DeepSeek token share: 22.6% (third place, within 2 points of Google at 24%)
- Open-weight share tripled since April (from ~11% to 29%)
- Token volume grew 29% MoM, spend grew 27% MoM, price per token remained flat
- Frontier closed-model prices up 12% per token in June, offsetting open-weight volume gains
- GLM 5.2 (Z.ai open-weight model) saw 50x daily volume growth, hit #11 ranking in 2 weeks
- Anthropic: 61% of spend on 32% of tokens; 72%+ of high-stakes use-case spend (coding agents, back-office)
- OpenAI: token share fell from 12.5% to 10.3%; cost per token up 50% relative to market in one month
- 1 in 8 enterprise customers now running open-weight models in production
- Claude Fable 5: reached 22% of Opus 4.8 request volume in 4 days before US export controls suspended access
- Image generation: OpenAI 53%, Google 39% (duopoly, no other model >5%)
- Video spend: Chinese labs 67% (ByteDance Seedance 49%, Kling, Alibaba Wan); xAI Grok Imagine 42% volume on 19% spend
The hook
Open-weight models hit 29% of production volume. DeepSeek is now closing in on Google. Here's what enterprise routing data reveals about the real AI hierarchy.
AI Gateway Production Index — July 2026
Every month, routes tens of trillions of tokens between production applications and AI labs, giving us a view of what AI usage actually looks like in today’s enterprise. We publish that view here. See the Production Index reports published in and .AI GatewayMayJune
The July index reports on AI Gateway data collected in June 2026.
In June, token volume and spend grew at nearly the same rate, 29% and 27%, while the price per token remained steady. The month before, spend had grown more than twice as fast as volume, driving the average token cost up by almost 20%.
The market spent more overall, but not more per token. Since April, open-weight models have climbed from a ninth of all token volume to nearly a third, at about a tenth of the average token price on the gateway. That alone should have pulled the price per token down. But as cheap volume rose in June, so too did closed-weight frontier prices, up about 12% per token. They offset each other, so the average price per token was flat.
This is evidence of the routing discipline June's report documented, now visible in the aggregate: high-volume work goes to low-cost models, high-risk work stays on the frontier. Investment in AI is still growing quickly, and companies are getting more strategic about balancing that spend across models.
Open-weight models ran 29% of gateway tokens in June on just under 4% of spend. That is nearly a third of the tokens for one twenty-fifth of the dollars.
Open-weight token share has almost tripled since April. Most of that volume comes from the continued growth of DeepSeek, now the third-largest source of tokens on the gateway at 22.6%, behind Anthropic and Google. Google's share of token volume slipped to 24% as its April surge unwound, leaving DeepSeek within two points of second place.
Roughly one in eight enterprise customers now runs an open-weight model in production. And on current trajectories, an open-weight lab will soon be the se...