Do you think AI companies are holding back much stronger models?

It’s possible a company has a model it hasn’t released, but that alone doesn’t show it’s sitting on a dramatically better one. Development, testing, and public release don’t happen on the same schedule, and outsiders usually can’t see the full picture.

A delay could reflect reliability, cost, safety checks, or product decisions. Without solid evidence, I’d treat the “they’re hiding the real thing” idea as a possibility, not a conclusion.

There’s a pretty ordinary explanation for staggered launches: a model can look impressive in testing and still be too expensive, inconsistent, or hard to serve at scale. Companies may also choose to release improvements in stages rather than put every change out at once.

That doesn’t rule out private experiments with more capable systems. It just means release timing by itself isn’t enough to tell us what those systems can do.

Maybe, but I wouldn’t assume there’s a secret finished super-model waiting in a cupboard. The public sees product announcements, not every internal experiment, so it’s easy to read a coordinated pattern into a few closely timed releases.

I’d watch for repeatable, independently checkable evidence of capabilities, not just the impression that one company answered another. Progress can look sudden in hindsight without proving anyone deliberately held back a breakthrough.