TypeSafe Raises $870M at $7.5B Valuation for Structured AI Model Jev

TypeSafe Inc., creator of the structured artificial intelligence model Jev, has raised $870 million in a funding round that values the company at $7.5 billion, as reported by SiliconANGLE. The financing was led by Andreessen Horowitz, with participation from Sequoia Capital, DCVC, and unnamed angel investors.

The capital infusion arrives less than a month after TypeSafe officially launched Jev. The company reports that the model has already been adopted by approximately one-third of the Fortune 500.

Traditional enterprise workflows using large language models typically require downstream software to parse and format unstructured natural language output before applications can consume it. Jev bypasses this post-processing requirement by directly generating structured output. The model supports three types of operations: returning a binary yes-or-no answer, selecting an option from a user-provided list, or generating a customized numerical score, such as ranking cybersecurity threat severity or support ticket priority.

To help applications mitigate hallucinations, Jev provides an explicit numerical confidence score alongside its list selections and numerical ratings. TypeSafe built the model on a custom architecture and trained it using a variation of reinforcement learning called reinforcement learning for calibrated decisions.

TypeSafe claims that its architecture enables Jev to process requests in under 700 milliseconds, asserting the model is up to 200 times faster and up to 100 times more cost-efficient than certain frontier LLMs. The startup plans to use the proceeds to expand its planned System One model family and build additional enterprise-focused tooling.