AIToday
Large Language ModelsAI Coding AssistantsZenn AI/MLPublished: Oct 11, 2026, 10:00 JST

TypeSafe AI's Jev opens to all, returns code-ready answers

TypeSafe AI's Jev opens to all, returns code-ready answers

TypeSafe AI's Jev, announced September 15, removed its waitlist on September 21, 2026, letting anyone register and try it.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

Jev comes from TypeSafe AI, which announced the new model on September 15. By September 21, 2026, the waitlist had been removed, making it open to anyone. The author had been watching the earlier announcement and tried it once registration opened.

What sets Jev apart is that it takes a state and a typed question, then returns a structured answer that code can use directly. Traditional language models produce natural language, which means a program has to extract the result afterward. Jev skips that step by returning only answers with a fixed type, and it can also output probabilities and confidence values between 0 and 1 per predefined option.

The author tested it in TypeSafe AI's playground. In the example, the Choice pattern was used to check the urgency of a question, with simple criteria stated in the question. The response's probability of true was 100%. The author frames Jev less as a chat or idea-review AI and more as a smart conditional branch inside a system, handling meaning-based routing and scoring.

FAQ
What can Jev actually do?
It offers three patterns: Noul returns a Yes probability between 0 and 1; Choice picks one from a set of known unordered options; Score measures a degree, such as severity, dissatisfaction or skill level, on a defined scale.
Does Jev need fine-tuning or keyword rules?
No. It uses zero-shot, meaning-based condition judgment, relying on the language understanding of a large language model rather than pre-trained rules or keywords.
What is Jev not good at?
It does not produce free-form sentences. It is suited to work where output candidates or scales can be defined in advance.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleByte Transformers beat subword models as scale grows