TypeSafe Inc., the developer of the Jev artificial intelligence model, has secured $870 million in a new funding round. This investment values the company at $7.5 billion. The round was led by Andreessen Horowitz, with participation from Sequoia Capital, DCVC, and several angel investors.
The Jev model, launched less than a month ago, has already been adopted by approximately one-third of Fortune 500 companies. Unlike standard large language models that generate natural language, Jev produces structured output. This allows applications to bypass additional reformatting steps, which accelerates software development and reduces coding errors.
Jev is designed to handle three specific request types: binary yes-or-no answers, list selection, and scoring tasks. The model also provides a confidence score with its output to help mitigate hallucinations. TypeSafe reports that Jev processes requests in under 700 milliseconds, claiming it is up to 200 times faster and 100 times more cost-efficient than some frontier models.
The company developed Jev using a proprietary training method called reinforcement learning for calibrated decisions and a custom model architecture. Jev serves as the first entry in the planned System One model series. TypeSafe intends to use the new capital to expand this product lineup and introduce additional enterprise-focused features.
