Glonce

Jev reaches 35% of Vercel's paid teams in two weeks, Vercel data shows

modelsThe Glonce editorsTechCrunch🔥56Glonce rating56Impact40Novelty45Relevance95Surprise40Players30Credibility65
Jev reaches 35% of Vercel's paid teams in two weeks, Vercel data shows

Jev, a lightweight decision model from TypeSafe AI, has become the fastest adopted AI model ever, at least by some measures, according to TechCrunch's reading of new Vercel data. The data, published on Thursday, shows that within two weeks of its launch Jev was being used by 35% of paid teams. That is nearly double the reach of the second most popular model, Sonnet 5.5. By Vercel's data it is the fastest adoption a model has ever seen in its first two weeks.

Jev was published in limited release on September 15. It is designed as a lightweight tool for choosing between a few predetermined outputs. Limiting the outputs lets it run cheaper and faster than alternatives like Sonnet, which makes it well suited to classifying large quantities of data quickly. The approach drew significant buzz in the weeks after release and inspired competing offerings from OpenAI and Amazon, among others, but until now there was little hard data to back up the good vibes.

The numbers come from Vercel's AI gateway, a unified API service similar to OpenRouter. TechCrunch calls it a necessarily limited view, particularly at risk of undercounting major labs such as OpenAI and Anthropic that offer their own APIs. Even so, it shows the buzz around Jev translated into real usage. Vercel itself was an early adopter: it uses Jev to replace GPT-6 Luna for classifying harmful commands.

The report also covers open source models. They now account for more than 70% of the tokens processed through Vercel's gateway and nearly 25% of the total spend. At the same time the market is growing fast: spending on frontier labs has grown tenfold, even as it makes up a smaller share of overall spend. TechCrunch says this gives credence to the idea that frontier models and open source alternatives may turn out to be complementary rather than competitive.

Key facts

  • 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.
  • Vercel's data puts this as the fastest first-two-weeks adoption any model has seen; TechCrunch qualifies the 'fastest ever' label with 'at least by some measures'.
  • Jev, from TypeSafe AI, was released in limited form on September 15 and picks between a few predetermined outputs, which makes it cheaper and faster than models like Sonnet for classification.
  • Open source models account for more than 70% of tokens and nearly 25% of total spend on the gateway; frontier-lab spending has grown tenfold while its share of spend has shrunk.
  • The gateway is a limited view that may undercount labs like OpenAI and Anthropic that offer their own APIs.

Why it matters

Until now Jev's reputation rested on buzz, and the article notes there was little hard data behind it. Vercel's numbers are the first usage evidence: 35% of paid teams in two weeks, with nearly double the reach of Sonnet 5.5, the second most popular model. The result also supports a design idea. A model that only chooses among a few predetermined outputs can run cheaper and faster than general models, and that approach has already prompted competing offerings from OpenAI and Amazon, among others. The wider report adds context: open source models carry more than 70% of tokens on the gateway, yet frontier-lab spending grew tenfold.

Who it affects

Teams that pay for Vercel's AI gateway are the population measured here, and 35% of them used Jev. The findings also matter to TypeSafe AI, Jev's creators, whom the article credits with a major accomplishment. OpenAI and Amazon are named as makers of competing offerings. Larger labs such as OpenAI and Anthropic are affected by the caveat that gateway data may undercount them, since they offer their own APIs.

How to use it

Jev is built for choosing between a few predetermined outputs, so the natural fit is classifying large quantities of data quickly. Vercel itself uses it to replace GPT-6 Luna for classifying harmful commands. The article does not say whether Jev is generally available or still in limited release, and it gives no price, size, architecture or benchmark results.

How solid is it

The adoption figures rest on Vercel's own AI gateway data, and the article describes the headline as true only 'at least by some measures', without specifying those measures. The article gives only percentages and ratios: no absolute counts of teams, tokens or dollars, and no share for Sonnet 5.5 beyond 'nearly double' in Jev's favor. It also quotes no one from Vercel or TypeSafe AI. TechCrunch itself says the data is a necessarily limited view.

Risks and caveats

The gateway is similar to OpenRouter and is particularly at risk of undercounting major labs like OpenAI and Anthropic that offer their own APIs, so the ranking among models may not reflect the wider market. Vercel was an early adopter of Jev, which is worth keeping in mind when reading Vercel's data. The period over which frontier-lab spending grew tenfold is not stated, and neither is the frontier labs' share of total spend. The idea that frontier and open source models are complementary is the article's interpretation, put as something that 'may' turn out to be true.