Vercel Reports Record‑Fast Adoption of Jev AI Model

Vercel’s internal analytics show that within two weeks of Jev’s launch, 35% of paid teams were using the model, nearly double the uptake of the next‑most‑popular option, Sonnet 5.5. This marks the fastest two‑week adoption ever recorded on the platform.

Jev, released by Typesafe AI on September 15, is a lightweight decision‑model that narrows its output set to a few pre‑defined choices. By limiting the search space, it runs faster and cheaper than larger models, making it suited for high‑volume data classification tasks.

The Vercel AI gateway data also reveal that open‑source models now handle over 70% of processed tokens and about a quarter of total spend, while spending on frontier labs has risen ten‑fold, suggesting a complementary relationship between open and closed‑source approaches.

Why it matters for GPU / AI infrastructure

For GPU‑centric workloads, a model that delivers low latency and reduced compute cost per inference can lower the total cost of ownership for AI pipelines. Jev’s efficiency profile means that inference servers can handle more requests per GPU hour, freeing up capacity for heavier training jobs or other latency‑sensitive services.