Jev becomes Vercel's fastest-adopted AI model
Original: Jev is Vercel’s fastest-adopted AI model ever
Why This Matters
Jev's rapid uptake signals real demand for cheap, fast classification models at scale.
TypeSafe AI's Jev decision model reached 35% of Vercel paid teams within two weeks of its Sept. 15 launch — nearly double the reach of runner-up Sonnet 5.5, per Vercel gateway data published Oct. 16.
Jev, a lightweight "decision model" from TypeSafe AI, has set an adoption record on Vercel's AI gateway, reaching 35% of paid teams within two weeks — the fastest uptake any model has seen in its first two weeks on the platform. That's almost double the penetration of Sonnet 5.5, the second most-used model.
Released in limited availability on September 15, Jev is built around a constrained output design: rather than generating open-ended responses, it selects among a small set of predetermined outputs. That constraint makes it faster and cheaper than general-purpose models, positioning it for high-volume classification tasks. Vercel itself replaced GPT-6 Luna with Jev to flag harmful commands.
The model's release sparked competing offerings from OpenAI and Amazon, but hard usage data had been scarce until now. Vercel's AI gateway — a unified API layer comparable to OpenRouter — provides the data, though it likely undercounts usage on labs' own APIs like Anthropic and OpenAI.
The report also tracks open source momentum: OSS models now handle over 70% of tokens through Vercel's gateway and nearly 25% of spend. Frontier lab spending has grown tenfold in absolute terms even as its share shrinks, suggesting open source and proprietary models may be growing in parallel rather than cannibalizing each other.