Twice as much as Sonnet 5.5: Jev became the most popular release on Vercel
Jev has become the fastest-adapting artificial intelligence model in history—at least by some measures.
Jev has become the fastest-adapting artificial intelligence model in history—at least by some measures.
According to data from Vercel published on Thursday, 35% of paid teams started using TypeSafe AI’s Jev model within the first 14 days of its release. That’s nearly double the reach of the second-most popular model, Sonnet 5.5, TechCrunch reports .
According to Vercel, this is the fastest rate of model adoption in the first two weeks in the platform's history and a significant success for its developers.
The Jev model was released into limited release on September 15. It is designed as a lightweight tool for choosing between multiple predefined response options. The limited amount of input data allows Jev to run much cheaper and faster than alternatives like Sonnet, making it ideal for rapid classification of large amounts of data.
This approach has generated a lot of buzz and inspired competitors—including OpenAI and Amazon—to create similar products, but until now, there has been a lack of real-world statistics to support its success.
The statistics were collected by Vercel AI Gateway (the only API service similar to OpenRouter). While these data are limited to a single platform and may not reflect the performance of large labs like OpenAI or Anthropic (which provide their own APIs), they clearly demonstrate that market interest has translated into actual usage.
By the way, Vercel itself became one of the first users of Jev, replacing the GPT-6 Luna model with it for classifying potentially malicious commands.
The Vercel report also highlights the rapid growth of open source models, which now account for over 70% of tokens processed through the Vercel gateway, and represent nearly 25% of total customer spend.
While open-source models are gaining more and more market share, the overall AI market is also growing rapidly. Spending on leading labs has increased 10-fold, even though their share of total spending has decreased. This supports the idea that advanced closed models and open alternatives may not compete, but complement each other.
Decision-making models like Jev are also vulnerable to prompt injections: what the study showed
Decision-making models like Jev are also vulnerable to prompt injections: what the study showed
Cloudflare has introduced the AI model Clef, a competitor to Jev. Unlike Jev, it can work with images and videos
Cloudflare has introduced the AI model Clef, a competitor to Jev. Unlike Jev, it can work with images and videos
Amazon has released its own analogue of Jev - an open decision-making model Strands Decider 2B
Amazon has released its own analogue of Jev - an open decision-making model Strands Decider 2B
“A useful classifier does not mean a revolution in AI yet.” How Ukrainian IT professionals evaluate the sensational Jev model — use cases in real products
“A useful classifier does not mean a revolution in AI yet.” How Ukrainian IT professionals evaluate the sensational Jev model — use cases in real products
Read the country's main IT news in our Telegram
Read the country's main IT news in our Telegram
