TypeSafe AI has raised $870M at a $7.5B valuation, a round led by Andreessen Horowitz with Sequoia and returning backer DCVC joining in. The company’s Jev model shipped on September 15 and went viral almost immediately.
Jev is not a language model. It runs on a transformer backbone but outputs probabilities rather than text, a format TypeSafe calls calibrated decisions. The pitch is speed and thrift: the company claims the model runs far faster than large language models and burns a fraction of the tokens, which suits automation work rather than writing.
TypeSafe says a third of Fortune 500 companies are already using it, an adoption curve that would be extraordinary for a product this young. Co-founder Diogo Almeida, a former OpenAI researcher, has argued that four years of progress on human language has done little for automation because machines speak a different language.
Almeida founded the company in 2024 with Sasha Sheng, previously a research engineer at Meta, and entrepreneur Erik Gafni. The round continues a pattern this year of capital chasing models that promise cheap, structured outputs instead of fluent prose.
