A new AI model called Jev returns a probability score instead of text. Its maker says that makes sorting tasks far cheaper, though the claim needs checking.
A new AI model called Jev does not write answers. It returns a number between 0 and 1. A score of 0.9 means a 90% chance the answer is yes. Its maker, TypeSafe AI, says that approach makes it about 100 times cheaper than comparable models for sorting tasks.
You ask a yes or no question, such as "Is this customer angry?" Jev replies with a score like 0.9. Software can act on that number without a person reading anything. TypeSafe built Jev for automation and classification, which means sorting items such as emails or reviews into groups.
TypeSafe launched Jev on September 18. Founder Diogo Almeida started the company two years ago after working on ChatGPT at OpenAI. He concluded that chatbots are built for human-readable language, while computers need clear, structured results.
Chatbots mostly charge for the text they write, counted in tokens, which are small chunks of words. Jev outputs a number, so it charges by the billion input tokens instead. TypeSafe also says Jev cannot make things up, because it produces numbers rather than sentences. That is the company's claim. A number can still be wrong.
These are individual tests. The 10 to 20 times figure from Mudholkar is well below the 100 times the company claims.
Jev targets businesses that run high volumes of repetitive checks, like flagging angry customers or routing emails. Costs matter there. The report notes a 625x price gap between the cheapest usable option, at $0.04 per million output tokens, and top models at $25. It also cites RaftLabs, which puts custom AI automation at $30,000 to $120,000 for 3 to 10 workflows, plus $500 to $8,000 a month to run.
The report does not give Jev's actual price list. It also offers no independent testing across many kinds of work. TypeSafe says Jev was trained only on synthetic data, meaning examples generated by computers. How that affects accuracy on your own messages is untested. Any business tempted by the savings should try Jev on its own data first.
Jev scores yes or no questions with a probability and may cut sorting costs, but test it on your own emails or reviews before trusting the savings.
EA has re-released 25 classic PC games at a discount until November 19, and Bethesda has finally given The Elder Scrolls VI a subtitle: Sentinel.
Read more: EA brings 25 classic PC games back, and Elder Scrolls VI gets a nameA user says Anthropic's Claude Code put his email in web requests, then scam messages arrived. The link is unproven, but it shows what AI tools can share.
Read more: User says Claude Code sent his email address to websitesOpenAI released a ChatGPT for analysts and bankers that shows where each number came from, a feature that could matter well beyond finance.
Read more: OpenAI launches a ChatGPT built for Wall Street analysts
