Jev is Vercel’s fastest-adopted AI model ever

TechCrunch reports Vercel data showing that within two weeks of launch, Jev was used by 35% of paid teams on Vercel’s AI gateway — nearly double the reach of the second most popular model, Sonnet 5.5, and the fastest first-two-weeks adoption Vercel has seen. TypeSafe AI published Jev in limited release on September 15 as a lightweight decision model for choosing among a few predetermined outputs, trading breadth for speed and cost versus models like Sonnet.

Vercel itself was an early adopter, replacing GPT-6 Luna for classifying harmful commands. The same report says open-source models now account for more than 70% of tokens through the gateway and nearly 25% of spend, while frontier-lab spending has grown tenfold even as its share of overall spend shrinks.

We’re treating gateway share as a real signal with known blind spots. Vercel undercounts labs that keep traffic on their own APIs — OpenAI and Anthropic especially — so this is not a global market crown. It is hard evidence that a narrow decision model can win where chat models are overkill.

That’s the useful AI story. Classification, routing, and “pick one of N” jobs don’t need a frontier essayist. They need calibrated, cheap, fast choices. If OpenAI and Amazon are already shipping competing decision offerings, as TechCrunch notes, the category is real.

The complementary market read also tracks: open weights take token volume; frontier labs still grow absolute spend. I’m watching whether Jev’s two-week spike holds past the novelty window, and whether “decision model” becomes a default line item next to chat and embed.

Jev’s constrained-output design is meant for high-volume classification. Competing decision-model efforts from larger labs followed the buzz around TypeSafe’s release.