Vercel Says New AI Model Jev Had Fastest Adoption On Record

Vercel says a new AI model called Jev became the fastest-adopted model it has ever tracked, reaching more than a third of its paying customers within two weeks of launch, according to company data published Thursday and reported by TechCrunch.

The data shows 35% of Vercel's paid teams were using Jev two weeks after it launched. That's nearly double the reach of the second-most popular model on the platform, Sonnet 5.5, in the same period, TechCrunch reported.

What is Jev and what does it do?

Jev is built by a company called TypeSafe AI. It launched in limited release on September 15, 2026. Unlike general-purpose chatbots, Jev is designed to do one narrow job: pick between a small, fixed set of possible answers. TechCrunch described it as a "decision model" rather than a conversational one.

Because Jev only has to choose from a short list of outputs instead of generating open-ended text, it can run faster and cheaper than models like Anthropic's Sonnet line. That makes it useful for tasks like sorting large batches of data into categories quickly, rather than writing essays or holding conversations.

How did Vercel measure the adoption numbers?

The figures come from Vercel's AI gateway, a service that routes API requests to different AI models for paying customers. TechCrunch compared it to OpenRouter, another unified API service that lets developers switch between model providers without rewriting their code each time.

Because the gateway only counts traffic that flows through Vercel's platform, the numbers have a built-in blind spot. Companies like OpenAI and Anthropic sell direct API access outside of gateways like Vercel's, so heavy users of those direct channels may not show up in the count. TechCrunch noted the data is "necessarily limited" and "particularly at risk of undercounting major labs like OpenAI and Anthropic."

Vercel itself was an early user of Jev. The company swapped the model in to replace GPT-6 Luna for the job of flagging harmful commands, according to TechCrunch's reporting.

What is the timeline from launch to this report?

• September 15, 2026: TypeSafe AI releases Jev in limited release.
• Following weeks: Jev draws attention in the AI industry, and competing decision-style models are announced by OpenAI and Amazon, per TechCrunch.
• Two weeks after launch: Vercel's gateway data shows Jev in use by 35% of paid teams, well ahead of Sonnet 5.5 in second place.
• October 16, 2026: Vercel publishes the usage data, and TechCrunch reports it as the fastest first-two-week adoption the company has recorded for any model.

Why does this matter beyond one company's chart?

Until this data came out, the attention around Jev was mostly talk. TechCrunch's report said there had been "little hard data to back up the good vibes" before Thursday. The Vercel numbers give the first concrete measure tying the buzz to actual usage inside paying customer accounts, even with the gateway's known gaps.

The release also sparked competition. TechCrunch reported that OpenAI and Amazon, among others, have rolled out competing offerings aimed at the same narrow, low-cost decision-making niche that Jev occupies.

What does the data say about open source models overall?

The same Vercel report found open source models now account for more than 70% of the tokens processed through its gateway, and close to 25% of total spending, TechCrunch reported. Tokens are the units AI systems use to measure how much text they process, so a high token share means open source models are handling a large volume of requests even if they aren't collecting the largest share of dollars spent.

At the same time, spending on frontier labs — the companies building the largest, most capable models — has grown tenfold even as those labs' share of overall spending shrinks, according to the data cited by TechCrunch. That combination suggests, in TechCrunch's framing, that frontier models and open source alternatives may end up complementing each other rather than replacing one another as the overall market for AI spending expands.

For businesses deciding which models to build on, the practical takeaway is narrower: a cheaper, narrowly scoped model aimed at simple classification tasks found faster uptake than general-purpose alternatives, at least among the customers running traffic through Vercel's gateway. Whether that holds once direct API usage from OpenAI and Anthropic is counted remains an open question the current data cannot answer.