Vercel published data indicating that Jev, a decision model from TypeSafe AI, reached 35% of paid teams on its platform within two weeks of a limited release on September 15. According to Vercel, that level of uptake is nearly twice the reach of the next most-used model, Sonnet 5.5, and represents the fastest adoption a model has seen in its first two weeks on Vercel’s gateway.
Model design and intended use
TypeSafe AI designed Jev as a lightweight decision model that selects from a small set of predetermined outputs. Limiting the possible outputs reduces compute cost and latency compared with more flexible models, making Jev suited to bulk classification tasks where many items must be labeled quickly.
The model’s narrower output space and lower cost-per-inference reportedly helped drive interest from developers and platform teams. Vercel said that the buzz around Jev has prompted competing offerings from vendors including OpenAI and Amazon.
What Vercel’s gateway data shows
The figures come from Vercel’s AI gateway, a unified API service the company compares to OpenRouter. Vercel acknowledged this is a limited view of the market and may undercount traffic that goes directly to providers such as OpenAI and Anthropic. Still, the company said the gateway data shows the early momentum for Jev translated into real usage on its platform.
Vercel also reported that it used Jev internally to replace GPT-6 Luna for classifying harmful commands.
Beyond Jev, Vercel’s report highlights a continued shift toward open source models: they now account for more than 70% of tokens processed through the gateway and nearly 25% of total spend. At the same time, spending on frontier labs has increased roughly tenfold, even as those labs represent a smaller share of overall spend—a pattern Vercel described as evidence that frontier and open source options may coexist as complementary parts of the market.
Original source: TechCrunch AI