It’s nice I love it but sometimes it just freezes like I send a message and it says responding to background activity?? I’m okay with that but it takes really long and pisses me off, second I don’t really like that if I don’t have the pro plan I can’t use it if I talked to it too much. So I hope you can fix the first one and maybe at least do something for the second one.

Their LLM machine is good ngl but they really need to actually make a better design of their app on macOS and iPadOS. It’s a terrible UX that compared to ChatGPT, Gemini, and especially Perplexity who’s the in terms of the app UX here. Claude app on iPadOS and macOS are really inflexible on the window sizing and it takes so much unnecessary space for nothing. On top of that, when I want to make a new chat on macOS, it will take me back to the desktop where the app is located instead of being floating like Gemini and ChatGPT which is really inconvenient to be a side assistant on my devices. I have given them about this feedback a while ago but nothing seemed to change which unfortunately I had to uninstall the app and replace it with Gemini and Perplexity on my iPadOS and macOS right now until they realize and fix this.

You greedy ghouls, remove your five-hour limits, you first attracted customers and we really believed you and went to you, you have an excellent artificial intelligence model, but it is impossible to work with it, remove the five-hour limits, leave the weekly ones so that a person can calmly choose when he can work once a week, personally, I completely moved from you again to ChatGPT and now I will recommend everyone to leave you because it is more or less seriously impossible to work with your model, and all models do about at the moment plus-minus the same thing, so all your efforts in attracting new customers and losing quality Old customers are simply disgusting, I hope soon you will receive feedback from your loyal customers in the form of unsubscribes from your subscriptions

Bugs:1. Artifacts popup: several stale / cached artifact entries even after the agent removing it many times and restarting the application many times. These artifacts are deleted and so don’t open/errs on tapping. 2. Transcripts are buggy and of poor quality overall for someone with a minor non-West accent. 30 to 50% or better voice-to-text transcripts in other competitive LLM chat interfaces. 3. Interactive bugs on text input. Complicated and many to even mention here4. Models are not intelligent at Auto judging thinking effort or level of analysis needed. Debatable and non-popular opinions and cases are concluded early with assumptions or fast responses even at high effort Opus with low analysis effort. Hard to get to surface questions.