Jev maker TypeSafe hits $7.5 billion as Microsoft matches its price with a Qwen-based rival

TypeSafe AI, the startup behind the decision model Jev, has raised $870 million at a $7.5 billion valuation in a round led by Andreessen Horowitz, with Sequoia and existing investor DCVC joining, TechCrunch reports. Jev came out on Sept. 15. TypeSafe says a third of Fortune 500 companies already use it.

Jev is built for software rather than people. A developer defines the possible answers in advance, a yes or no, one of several choices, or a score on a rubric, and Jev returns each with a probability attached. TypeSafe charges $0.042 per million input tokens, the chunks of text a model reads, and nothing for output, according to its launch post.

On Friday, the day the round was reported, Microsoft released Microsoft-Decision-1. The Register reports the first version is built on Qwen3.5-9B, an open-weight model from Alibaba’s cloud arm, and that Microsoft plans to rebase it on its own models and OpenAI’s. The price in Microsoft’s announcement is identical to Jev’s, down to free output.

The Register counts more than 100 decision models now chasing Jev, including OpenAI’s Decisions API, Cloudflare’s Clef, Liquid AI’s d1, and models from Snowflake, Perplexity and H2O.ai. Microsoft’s, Cloudflare’s and H2O.ai’s are built on Alibaba’s Qwen.

Classifiers that answer with probabilities are old; The Register notes machine-learning researchers have long been able to build them. What Jev changed is that a developer gets one from a prompt. Armin Ronacher, CTO of Earendil, told TechCrunch in September that competitors would spring up: “We should have seen this earlier in many ways, but presumably because the LLMs are so cheap and subsidized, you often don’t have to be creative yet.”

Cisco tested Jev against a 1-billion-parameter classifier it trained on its own 24-category safety rulebook. Held to a 0.5% false-positive budget, meaning it could wrongly flag no more than one safe conversation in 200, Cisco’s model caught 61% of the unsafe conversations in its multi-turn test set and Jev caught 34%, Cisco wrote. Jev ran close to a 31-billion-parameter language model doing the same job.

Microsoft matched Jev’s price in under a month. What TypeSafe keeps is the head start: the customers it says it has, and a training method built on synthetic data that co-founder Diogo Almeida told TechCrunch was “one of the best bets I’ve ever made in my life.”

Jev is named for William Stanley Jevons, the economist who noticed that making a resource cheaper tends to increase how much of it gets used. Almeida is betting cheap decisions end up inside every piece of software. That can be right while TypeSafe’s share of it shrinks, because Microsoft is already selling its version through Foundry, its AI platform for enterprise customers, and through OpenRouter.

Sources

• TechCrunch: The maker of non-text AI model Jev valued at $7.5B just weeks after launch
• TechCrunch: A new kind of AI model from a ChatGPT inventor is thrilling developers
• The Register: Microsoft leans on open weight model from Chinese AI lab to challenge Jev
• The Register: Cloudflare tries to outplay Jev with open-weight Clef models
• Microsoft: Microsoft-Decision-1 in Microsoft Foundry
• TypeSafe AI: Introducing System One Models & Jev
• Cisco: Jev decision models on trial