Hey gang, sorry for the mixup - there was a misunderstanding that 2.7 could fill the role in Assistant just as well based on how Moonshot was describing it, but we realized too late that it wasn't multimodal (etc).
Generally speaking, when models in the same class / use case get a clear cut upgrade, we will do so with less notice since the end result should just be higher quality interactions without any "breaking" changes in model behavior. But you're right that we should at least give some amount of notice, even if shorter than our usual "this model is going away forever and you should think about how to replace it because replacing it might come with significant UX changes", like a whole model family going away or a major version bump.
2.6 is back now, but I don't expect that 2.5 will be back, since it generally supersedes 2.5 for a modest price increase - and we prefer to keep the model list clean of too many options. (I also wouldn't be surprised if our upstream providers start dropping support for it as time goes on & people stop using 2.5 globally; we use primarily 3rd party hosts for privacy reasons and are at their whim if they decide to drop it - but that's just my own speculation)
But if we change our minds, well, you will see it in the list again 🙂
Thanks!
