Jev, Textless AI: TypeSafe AI Raises $870 Million

TypeSafe AI is betting on Jev, a model that does not produce text but rather qualified probabilities of "calibrated decisions." The company presents this approach as faster and more token-efficient than LLMs, focusing on automation. Following its launch on September 15, the company announced a funding round of $870 million, valuing it at $7.5 billion.

Jev Aims for Automation with Probabilities Instead of Text

Jev is based on a transformer architecture, but it is not a large language model and does not generate text. The model produces probabilities, which TypeSafe AI refers to as "calibrated decisions." According to the company, Jev operates significantly faster than LLMs and requires far fewer tokens. TypeSafe AI emphasizes that Jev is designed to automate tasks rather than generate text or code. Co-founder Diogo Almeida explains that human language performance is not relevant for automation, as computers use a different language.

A Team Founded in 2024 by Profiles from OpenAI and Meta

Diogo Almeida, a former researcher at OpenAI, is one of the co-founders of TypeSafe AI. The company was established in 2024 by Sasha Sheng, previously a research engineer at Meta, and Erik Gafni, an engineer and entrepreneur, alongside Almeida.

$870 Million Raised and Valuation at $7.5 Billion

TypeSafe AI raised $870 million in a funding round led by Andreessen Horowitz, with participation from Sequoia and existing investor DCVC. This round values the company at $7.5 billion. The launch of Jev took place on September 15, and its adoption has been rapid. According to TypeSafe AI, one-third of the companies listed in the Fortune 500 are already using this model.