They seem to have removed auto-correct. That is NOT an improvement, unfortunately.

Also the startup time / connection time is significantly longer than before. Once in, the basic IM function seems nice and snappy, but there are periods where it seems to “hang” while reconnecting. I’m not sure what the problem is. The previous version didn’t do that.

I’m also not crazy about the add-in fee for speech recognition, even though I do understand their desire to charge for a “premium” service, but I’d rather see it backed into the price of the app, which already isn’t cheap. At the very least, I’d like to see a trial period first to see how well it worked with my speech patterns before I spend money on the feature.