TypeSafe Launches Jev, a Decision-Only AI Model Claiming 444x Cost Savings Over LLMs

TypeSafe co-founder Diogo Almeida announced Jev on September 15, a specialized AI model designed exclusively for structured decision-making in automated workflows rather than text generation. The model accepts a structured state and a set of questions, returning typed outputs such as boolean, choice, or scored answers, but cannot produce free-form sentences. TypeSafe's own benchmark data shows Jev matching Claude Sonnet 5's accuracy at 67.8% across several workflows while costing roughly 294 times less per case and responding significantly faster. The company describes Jev as a 'System One Model' — borrowing from Kahneman's fast, intuitive thinking framework — and positions it as a cheaper replacement for frontier model calls that only route or classify within an agent pipeline. Vercel has already integrated Jev into its AI SDK, and direct API access from TypeSafe remains in early access.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in