Typesafe, the developer of AI model Jev, has raised $870 million at a valuation of $7.5 billion, SiliconANGLE reported on Thursday.

The report said the investment was led by Andreessen Horowitz. Sequoia Capital, DCVC and unnamed angel investors also participated. Typesafe secured the funding less than a month after releasing Jev. The company said about one-third of Fortune 500 companies have already adopted Jev.

When enterprise applications assign tasks to AI models, the models usually respond in sentences that people read. Applications must then extract only the needed values from those sentences to use them. Jev returns only values that applications can use immediately, such as yes-or-no answers, item numbers and scores. It does not require a processing step.

Jev handles three types of requests: yes-or-no answers, selecting an item from a list and scoring. Developers can set the scoring criteria. Examples include evaluating the severity of a cybersecurity alert or assigning a numerical urgency level to a customer support request.

Because the output is concise, developers can write less code for data processing. Development time falls and simpler code reduces the risk of errors. When Jev assigns a score or selects an item, it also provides a confidence figure for answer accuracy. Applications can use that figure to reduce damage from hallucinations.

Typesafe trained Jev using what it called reinforcement learning for calibrated decisions, a modified form of reinforcement learning. It also developed a new model architecture. The company said Jev processes requests within 700 milliseconds. It said Jev is up to 200 times faster than some frontier LLMs and up to 100 times more cost-efficient.

Jev is part of Typesafe's System One series of models that the company is preparing. Typesafe plans to use the investment to expand the series. It also plans to introduce enterprise features so large companies can use the models more easily.