Vercel reports Jev model reached 35% adoption among paid teams within two weeks of launch
Usage data from Vercel’s AI gateway shows Jev, a lightweight decision model from TypeSafe AI, outpaced Sonnet 5.5 in early adoption, while open source models accounted for over 70% of tokens processed.
1 source · cross-referenced
- Jev, a lightweight decision model from TypeSafe AI, was adopted by 35% of Vercel’s paid teams within two weeks of its September 15 launch.
- Vercel’s AI gateway data shows Jev’s adoption was nearly double that of the second most popular model, Sonnet 5.5.
- Open source models accounted for more than 70% of tokens processed through Vercel’s gateway and nearly 25% of total spend.
- Spending on frontier labs grew tenfold even as their share of overall spend declined, suggesting frontier models and open source may be complementary.
Vercel’s AI gateway, a unified API service, published data on Thursday indicating that Jev, a decision model from TypeSafe AI, became the fastest adopted AI model in its first two weeks. Within that period, Jev was used by 35% of Vercel’s paid teams, a share nearly double that of the next most popular model, Sonnet 5.5. Vercel described the adoption as the fastest ever recorded for a model in its first two weeks.
Jev was released in limited form on September 15 and is designed as a lightweight tool for selecting between a small set of predetermined outputs. By limiting the output options, Jev reduces computational cost and latency compared to larger general-purpose models, making it suitable for tasks like classifying large volumes of data quickly.
The adoption figures come from Vercel’s AI gateway, which aggregates usage across models accessed through its unified API. Vercel acknowledged that the data presents a necessarily limited view of the broader AI model market, noting that major labs such as OpenAI and Anthropic operate their own APIs and may not be fully represented in these metrics.
Despite the limited scope, Vercel’s data also showed strong momentum for open source models, which accounted for more than 70% of tokens processed through its gateway and nearly 25% of total spending. At the same time, spending on frontier labs grew tenfold, even as their share of overall spend declined. Vercel suggested this pattern indicates frontier models and open source alternatives may be complementary rather than purely competitive.
Vercel itself was an early adopter of Jev, using the model to replace GPT-6 Luna for classifying harmful commands in its systems.
- Oct 17, 2026 · The Verge — AI
ChatGPT Chrome extension gains tab access and sidebar controls
Trust79 - Oct 17, 2026 · TechCrunch — AI
Replit ends Anthropic partnership, cites cost and performance in shift to OpenAI
Trust74 - Oct 16, 2026 · TechCrunch — AI
Nvidia alum launches fund to back ‘physical AI’ startups in robotics, materials, and life sciences
Trust78