Table of Contents
In a new blog post, Vercel published that Jev, TypeSafe AI’s evaluation model for fast software decisions, saw the fastest first-day adoption of any model in AI Gateway history, reaching roughly 13% of paid AI Gateway teams during its first 24 hours on the platform.
Context: Gateway is Vercel’s unified API for routing requests across hundreds of AI models and providers, with centralized credentials, spend controls, logging, and automatic failover.
According to Vercel, this adoption rate was about twice the adoption of the GPT-5.6 family and about six times the adoption of Claude Fable 5.1 over the same first-day window.
Jev is not a conventional chat model. Built by TypeSafe AI, it is a “System One” model for structured decisions inside software workflows. Instead of generating free-form text token by token, it evaluates application state and returns typed outputs, such as choices, scores, and Boolean probabilities.
Furthermore, Jev is a probabilistic decision model for software, where “state goes in” and typed answers come out. The model can evaluate multiple declared questions in parallel, which Vercel says removes unnecessary text generation and makes it easier to automate clear cases while routing uncertain ones to review.
Bart Hillerich
Jev is available through Vercel AI Gateway under the model ID typesafe-ai/jev. Vercel lists the model as an evaluation model with pricing of $0.042 per 1 million input tokens, with no maximum output-token field because the model does not produce text in the same way as a standard language model.
