Jev Reaches 35% of Paid Vercel Teams in Record Two-Week Surge
Jev reached 35% of paid Vercel teams within two weeks, nearly doubling Sonnet 5.5, as open models passed 70% of gateway token volume in new usage data.
Summary
As of October 16, 2026, Vercel’s AI Gateway data showed Jev, Typesafe AI’s decision model, reached 35% of paid teams within two weeks of its limited September 15 launch. That was the fastest first two-week adoption Vercel has recorded and nearly twice Sonnet 5.5’s reach. Jev cheaply and quickly chooses among a few predetermined outputs, suiting high-volume classification; its debut prompted competing offerings from OpenAI and Amazon. Vercel replaced GPT-6 Luna with Jev for harmful-command classification.
The gateway, a unified API like OpenRouter, captures only part of the market and may undercount OpenAI and Anthropic, which offer their own APIs. Even so, it provides usage evidence behind Jev’s early attention. Open-source models now generate more than 70% of gateway tokens but nearly 25% of spending. Frontier-lab spending has increased tenfold despite losing overall spending share, suggesting both segments are expanding and may be complementary rather than purely competitive.
Positives
- 35% of paid Vercel teams used Jev within two weeks, the fastest early adoption of any model measured by the company.
- Jev achieved nearly twice the reach of Sonnet 5.5 among paid teams during the measured period.
- Vercel replaced GPT-6 Luna with Jev for classifying harmful commands after becoming an early adopter.
- More than 70% of tokens processed through Vercel’s gateway now come from open-source models.
- Frontier-lab spending grew tenfold even as open-source models gained market share.
Risks & concerns
- Vercel’s AI Gateway covers only part of the market, limiting conclusions about broader model adoption.
- OpenAI and Anthropic may be undercounted because both companies provide customers with their own APIs.
- Jev chooses only among a few predetermined outputs, restricting it to constrained decision and classification tasks.
- Frontier labs account for a shrinking share of overall spending despite their tenfold spending growth.