The key facts

  • Jev (Typesafe AI) reached 35% of Vercel's paid teams in two weeks — twice as much as Sonnet 5.5
  • A classification model with a limited set of outputs, cheaper and faster than alternatives
  • Open-source models handle over 70% of gateway tokens, but spending on frontier labs grew tenfold

Record-breaking adoption speed

Data published by Vercel on Thursday shows that the Jev model was used by 35% of the platform's paid teams within two weeks of its launch — nearly twice as much as the second most popular model, Sonnet 5.5. According to Vercel, this is the fastest model adoption in the first two weeks in the gateway's history. The model was created by Typesafe AI and was released in a limited release on September 15.

What is Jev

Jev is designed as a lightweight tool for choosing between several predefined outputs. The limited set of answers allows the model to work cheaper and faster than alternatives like Sonnet, making it suitable for fast classification of large volumes of data. The launch generated significant buzz and pushed competitors, including OpenAI and Amazon, to create similar offerings, but until now there has been little objective usage data.

Data limitations

The statistics come from Vercel's AI gateway — a unified API service similar to OpenRouter. This is a limited slice of the market: it very likely underestimates large labs like OpenAI and Anthropic, which offer their own APIs. Nevertheless, the data shows that the hype around Jev has translated into real usage. Vercel itself became an early user of the model, replacing GPT-6 Luna for classifying malicious commands.

Open-source and frontier models

The report also notes the continued growth of open-source models: they account for more than 70% of tokens processed through the Vercel gateway and nearly 25% of total spending. Meanwhile, the market is growing: spending on frontier labs grew tenfold, although their share of total spending is declining. This is consistent with the idea that frontier models and open-source alternatives may turn out to be complementary rather than competing.