TypeSafeAI, a startup founded by former OpenAI researcher Diogo Almeida, has opened access to Jev, removed its waitlist and offered each new user $5 in credits. Unveiled on Sept. 15, Jev is designed for rapid judgments rather than text or code generation: it classifies content, assigns scores and estimates whether conditions are met. The company calls it a System One Model, drawing on the concept of fast, intuitive decision-making. In an internal test, Jev completed 27 judgment tasks in 0.114 seconds at a cost of $0.0001, compared with 8.566 seconds and $0.0139 for an OpenAI model. TypeSafeAI says that makes Jev about 75 times faster and one-170th the cost, though the figures are self-reported and require external verification. Jev accepts factual and closed-ended questions and returns probability estimates within 70 to 500 milliseconds; input costs $0.042 per million tokens and output is free. The company says developers can adapt it to multiple tasks by changing judgment criteria in natural language rather than conducting separate training. In a Bitcoin price test, Jev assigned a 55% probability to a Binance one-minute candlestick reaching $85,000 before 2027, versus 48% for $90,000 and 34% for $100,000, while Polymarket's corresponding probability for $85,000 was 81%. On Sept. 16, Monad Chief AI Engineer Jarrod Watts connected Jev to price data and built a trading bot that submits limit orders to the Kuru chain order book roughly every 300 milliseconds. The bot was open-sourced on GitHub as jarrodwatts/jev-trader.