TypeSafe AI, the San Francisco startup behind the viral non-text AI model Jev, announced on Friday that it has raised $870 million at a $7.5 billion valuation. The Series A round was led by Andreessen Horowitz, with participation from Sequoia Capital, existing investor DCVC and other angel investors.
The deal, confirmed by TypeSafe's legal counsel Wilson Sonsini and first reported by Bloomberg, lands less than four weeks after Jev's September 15 launch. It ranks among the largest Series A rounds in startup history — the kind of check usually reserved for far older companies.
A model that decides instead of talking
Jev is built on a transformer architecture, but it is not a large language model. Instead of generating text, it produces what TypeSafe calls "calibrated decisions" — probabilities and constrained answers such as a choice, score or classification that software can act on directly.
The pitch is speed and cost for automation. TypeSafe says Jev handles the structured decisions buried inside software — whether to escalate a support ticket, whether an image passes a content policy, whether a search result is spam — far faster and with far fewer tokens than routing every decision through a chatty language model.
Co-founder Diogo Almeida, a former OpenAI researcher who worked on InstructGPT, has argued that computers speak a different language than humans, and models built for automation should reflect that.
TypeSafe calls Jev a "System One Model," trained with what it describes as Reinforcement Learning for Calibrated Decisions. The company claims the approach makes Jev roughly a hundred times cheaper and faster than conventional language models at decision tasks.
Enterprise adoption at unusual speed
Investors are betting on traction that arrived almost instantly. TypeSafe says about a third of Fortune 500 companies are already using Jev, and the model surpassed one million users shortly after its launch video went viral. On launch day the API briefly buckled under demand — an early signal of developer appetite.
The round caps a dizzying first month. TypeSafe emerged from stealth on September 15 with a $40 million seed round led by DCVC. Weeks later, The Information reported the company was in talks to raise $1 billion or more at valuations of at least $10 billion.
The final terms — $870 million at $7.5 billion — came in lower than those early conversations, but the valuation still places a month-old product among the most valuable AI startups in the world.
TypeSafe was founded in 2024 by Almeida alongside former Meta research engineer Sasha Sheng and engineer-entrepreneur Erik Gafni.
Skeptics want proof
Not everyone is convinced the bet pays off. The core question is calibration: whether Jev's confidence scores stay reliable once deployed in real customer workflows, not just demos. Large language models have also grown steadily cheaper and faster, which could compress the model's cost advantage as the big labs keep optimizing inference.
For now, though, the money is talking. With Andreessen Horowitz and Sequoia both on the cap table, TypeSafe has the war chest to find out whether decision-first models become a new category of machine intelligence — or a fast, cheap footnote in the LLM era.
By Hamza Faisal
By Hamza Faisal
