To claim that at the top of your description is a bit incorrect. More times than I can count the two different models I have used have simple made things up to give me the answer it thought I wanted. I have wasted 3 days trying to fix a problem with an RP system prompt only to finally force the model to admit there was nothing wrong with the promot in the first place it was simply ignoring my instructions because it was a bit overloaded. I now load a set of Directives into every chat window. One demands honesty about its failures to perform as requested. I have to tell it not to lie.To whoever is responsible for training these models this behavior is unacceptable. I should not have to demand honesty. I should not waste my time and wreck a perfectly good system prompt with a bunch of 'fixes' that cannot possibly work because the failure is not my work but rather the model failing. And then lying about it. And it is a lie not a delusion. Because when pushed hard enough it will admit it knew all along but still wasted my time.

I really really like how it combines all the best AI tools into one three to try or pro subscription for $18 a month at the time of this review. I think there's some issues with the pro version as I could only get three videos for the $18 before it asked for more money. Asking or talking with the agent AI agent about this problem it tried to get some work around lower price 720 pixel resolution, but seems like the credit just burned up really quickly. It only took a couple of days to go through a few chats and three videos. That said, the system itself almost works there's one small thing where I would like to listen to the responses and there is a little speaker button, but 99 terms out of 100 a little button won't playback the text And the voice that it reads back and it's just robotic and not quite a storyteller voice. So just a couple of issues and I hope they continue to work on the product and I'll pay again for another few months without any improvement or development I will.. UPDATE 7/23, now it’s got unlimited chat, which is what the deal advertised. I’m concerned when I see someone stuck “in an endless loop” unable to cancel their subscription. $18 a month is not a nice leak, but hopefully Venice can make everyone happy.

I don’t know if the developers actually read their reviews. But I have a small request. I downloaded this app because of all the great reviews and amazing talk i’d heard about it online. But unfortunately, as soon as I launched the app itself, I couldn’t use it with my screen reader. I’m visually impaired and use VoiceOver on my iPhone to do absolutely everything. And even though it is great and it helps me with almost anything, it doesn’t seem to work with this app. I won’t pretend to know how app development works, but I would like to request if possible, for Accesibility to be taken under consideration for any future updates of this app. Please and thank you. I’d love to be a contributor to the app itself and get all the benefits from it that I’ve heard so many people rave about. And the only way I will be able to do that is with a few changes. thanks again for taking my suggestion into consideration.

Hello, everything is good at this app, i used DeepSeek for a really long time and DeepSeek’s AI and Censorship didnt really let me ask some questions and your app fascinated me due to almost absent censorship and it letting me ask some stuff. But please, there’s a issue i found. Its a literal predicament. If it will not be fixed, it will be a active problem to not just me, but other people too. Basically this problem is, i randomly got a message from your AI after i tried to ask a simple question. Really. The question doesnt really matter too. If you even say ‘hi’ or like ask it very normal stuff, like things about animals and things, man, even send a dot, the AI literally doesnt answer. It gives me the ‘a unknown error occurred. Please try again.’ Error. And this error lasted for 4 hours and it is still lasting. Please, i dont know what to do. Fix this ASAP. The AI does not work.