Vercel says open-weight models reached 62% of AI Gateway traffic in August
Key Takeaways
- •Open-weight models rose from 11% of Vercel’s token volume in April to 62% by August 22.
- •Vercel said the figures come from its AI Gateway and reflect production workloads from real companies.
- •Open-weight models can cost about one-tenth as much as closed-source alternatives, helping explain the shift in traffic.
- •Anthropic’s closed models still accounted for 61% to 65% of spending on the gateway in recent periods.
- •DeepSeek has overtaken Google in Vercel’s token processing share.
Two months ago, open-weight AI models were a minority player on Vercel’s infrastructure. Now they are leading by volume.
Vercel CEO Guillermo Rauch said on August 22 that open-weight models accounted for 62% of all tokens processed through Vercel’s AI Gateway, up from 28.4% on June 24.
Rapid growth in token share
Open-weight models accounted for just 11% of Vercel’s token volume in April. By June, that figure had risen to 29%. By late August, it had crossed 62%.
Vercel’s AI Gateway serves as a routing and traffic management layer for AI-powered applications, so the figures reflect production workloads from real companies rather than benchmark tests or lab conditions.
The shift matters because token share and dollar spend do not move in lockstep. Open-weight models can run at roughly one-tenth the price of their closed-source counterparts, which helps explain why enterprises may route larger volumes of routine or high-throughput tasks to them while still reserving premium models for workloads where they are worth the extra cost.
Spending remains concentrated
Even with open-weight models accounting for 62% of token volume, they do not represent 62% of spending.
Anthropic’s closed models captured between 61% and 65% of total expenditure on the gateway in recent reporting periods. Claude is processing a minority of tokens while generating a majority of the revenue.
DeepSeek has risen to the top, or near the top, of Vercel’s token volume leaderboard, overtaking Google in processing share.
Broader enterprise deployment trend
Major companies including AT&T and Coinbase have been noted among those emphasizing cost-reduction strategies in AI deployment, which aligns with the workload routing patterns reflected in Vercel’s data. For developers and platform operators, the trend suggests that model selection is becoming increasingly workload-specific: volume can shift toward cheaper open-weight systems even as spending stays concentrated in a smaller set of higher-priced models.
