Jev Fit

It depends on the mode, and that is the first thing to check. In demo mode the page states answers are seeded random numbers shaped like the real API response, not Jev, so a score obtained there says nothing about the model. The site also states it is not affiliated with TypeSafe AI, so treat it as a third-party tool.

It rejects anything that does not read as a software task, with the guidance to describe an input, a decision and what the program does with it. The field caps at 5,000 characters. The source does not document how strict that filter is, what happens between rejection and a verdict, or any calibration behind the success potential number.

There is a public endpoint, a single POST /api/fit with an idea field, shown right on the page. The source does not document authentication, rate limits or pricing for it, so treat it as a hobby API and check before depending on it in anything with traffic.

The page itself warns against pasting personal data or secrets, which is the right instinct when there is no account. Results can be turned into a shareable link, posted to X or deleted, so anything you paste becomes shareable if you use those buttons. A privacy page is linked but no retention window for the idea text is stated.

Ideas that do not read as a software task get a 'Not evaluated' verdict with the guidance to describe an input, a decision and what the program does with it. The idea field caps at 5,000 characters.

No account is involved. After evaluating, you can copy a link, post the result on X, or delete it with a Yes/No confirmation, and the page carries a 'Buy me a coffee' link. Because there is no login, the page warns against pasting personal data or secrets.

No calibration is documented. The page presents success potential on Jev but does not state how it is computed, and in demo mode the number is a seeded random value shaped like the real API response, so read it as a directional fit indicator rather than a probability.