Jev is Vercel’s fastest-adopted AI model ever
TechCrunch Russell Brandom ● Covered by 3 sources
Vercel says Jev is its fastest-adopted AI model ever, with 35% of paid teams using it in two weeks. That’s a loud sign that cheap, narrow models can catch on fast when they actually save work.
Based on reporting by TechCrunch, Russell Brandom — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Jev has done something a lot of buzzy AI launches only promise: it’s getting used. Vercel’s data, published Thursday, says the decision model reached 35% of paid teams within two weeks of launch, with nearly twice the reach of Sonnet 5.5, the next closest model in the same data set.
That makes Jev the fastest-adopted model Vercel has seen in its first two weeks. For TypeSafe AI, the company behind it, that’s a serious early win. The model was released in limited form on September 15 and is built for a very specific job: picking between a few fixed outputs.
That constraint is the trick. By keeping the choice set small, Jev can run cheaper and faster than larger alternatives like Sonnet, which makes it useful for classification work where speed and volume matter more than creative range. The pitch is simple, and apparently persuasive.
Vercel’s numbers come from its AI gateway, a shared API layer similar to OpenRouter, so they don’t capture everything. The company itself warns that major labs with their own APIs, including OpenAI and Anthropic, are likely undercounted. Even so, the report says the buzz around Jev turned into real usage, not just demo-room applause.
Vercel was also an early Jev customer, using it instead of GPT-6 Luna for classifying harmful commands. And the broader picture in the report is hard to miss: open source models now make up more than 70% of tokens flowing through Vercel’s gateway and almost 25% of total spend, while frontier lab spending has grown tenfold. The market isn’t picking one side. It’s spreading out.
My take — AI-written commentary, not fact-checked reporting
The neat little lesson here is that AI buyers are often less impressed by grandeur than by models that do one boring thing well and cheaply. The industry keeps selling giant brains; the money keeps drifting toward tools that behave like reliable clerks. That’s not a failure of ambition, just a reminder that most production work is not a poetry contest.
Read more about this at: TechCrunch
Related stories
Jev, an AI for making quick decisions, has been a viral hit. But OpenAI is hot on its heels
Fortune ·
15