With Prism 8.2, we introduced a new concept: Prism Labs. Features released as part of Prism Labs are new, and we welcome feedback to make them better.
Prism Labs features are fresh from the GraphPad bakery. We think they are fully baked, but label them as a "Labs" because there is some chance they'll turn out to be only half-baked. These features represent our latest and most innovative features, and while they've been extensively tested, they can still be considered "experimental".
The first (and only as of 8.2) Prism Labs feature is a new way to detect "bad" fits during nonlinear regression. The new method detects "unstable" parameters rather than "ambiguous" fits. It seems to work better. Choose it on the Confidence tab of the nonlinear regression dialog.