Vercel published data showing that within two weeks of launch, Jev was being used by 35% of paid teams on its AI gateway — nearly double the reach of the second most popular model, Sonnet 5.5.
Jev, from TypeSafe AI, was published in limited release on September 15. It is a decision model: instead of generating free text, it chooses between a small set of predetermined outputs. Restricting the output space is what makes it cheaper and faster than general models, which suits classifying large volumes of data.
The measurement has limits worth naming. Vercel's gateway is one API service, so it is likely to undercount labs that offer their own APIs directly. But it does show the attention around Jev translated into real usage — Vercel itself adopted it early, replacing GPT-6 Luna for classifying harmful commands.
The same report has a second finding: open source models now account for more than 70% of tokens processed through the gateway, and nearly 25% of total spend. Meanwhile spending on frontier labs grew tenfold even as their share shrank.
Why it matters
- A narrow model that only picks between options can beat a general one on cost and latency for classification work. That is a design choice worth copying.
- The open-source share numbers suggest the two tiers are complementing each other rather than one replacing the other.
- Adoption data from a single gateway is directional, not a market census. Treat it as a signal, not a scoreboard.