Typesafe AI launched Jev, a specialized artificial intelligence model designed for cryptocurrency applications, to public availability on Sunday after two years of development. The company raised $40 million and provided new users with $5 in credits upon release.
Unlike conversational AI systems, Jev operates through binary or multiple-choice questions and returns confidence scores within 70 to 500 milliseconds. The model costs $0.042 per million input tokens with no output fees. By design, Jev cannot generate text, which developers say eliminates the generation of false information while enabling integration into automated systems.
Early Applications in Price Prediction
Bitcoin.com News tested Jev against data from Polymarket, a prediction market tracking bitcoin price targets through 2026. When asked whether bitcoin would reach $85,000 by the end of 2026, Jev assigned a 55% probability while Polymarket traders showed 81% confidence. For $90,000, Jev scored 48%; for $100,000, it scored 34%. On a potential dip to $70,000, Jev returned 50%.
Trading Bot Development
Jarrod Watts, lead AI engineer at Monad, built an onchain trading bot using Jev that executes trades approximately every 300 milliseconds. The bot receives price feed data, receives a buy or sell signal from Jev, and places post-only limit orders on Kuru's onchain order book. Watts open-sourced the bot as jarrodwatts/jev-trader.
Multiple cryptocurrency projects have begun experimenting with Jev for applications including stablecoin strategies, token scoring systems, and meme coin launch analysis. Developers are creating templates and forks across the ecosystem as adoption accelerates.
The shift toward Jev reflects growing interest in bounded, probabilistic AI systems over conversational interfaces for financial decision-making infrastructure.
