I can’t even get my first chat to fire up.
I’m using the app during the free trial, but I can’t get it to switch models or even get my first chat going. I can’t justify subscribing to an app that doesn’t work, or even justify using it for free.
Unstable and unusable, clearly vibe coded without testing
Ever since I started using this app months ago, there have been nothing but bugs with basic functionality. It seems vibe coded and untested. Frankly, this is not something that should be acceptable as a paid, public release on the App Store.
For example, every time the app is opened, or a new chat is started, it immediately reloads the most recent chat, so it is impossible to start a new chat without it loading the old one.
Another example: Grok never worked despite having explicit support in the app because LLMConnect was trying to include parameters that don’t exist in its requests. These both seem like things that would come up very easily if the app was tested properly, even by an individual.
Again this kind of thing should be kept in beta until it’s actually functional as an MVP. It seems like the developer blatantly ignored best practices just to collect a subscription fee.
For example, every time the app is opened, or a new chat is started, it immediately reloads the most recent chat, so it is impossible to start a new chat without it loading the old one.
Another example: Grok never worked despite having explicit support in the app because LLMConnect was trying to include parameters that don’t exist in its requests. These both seem like things that would come up very easily if the app was tested properly, even by an individual.
Again this kind of thing should be kept in beta until it’s actually functional as an MVP. It seems like the developer blatantly ignored best practices just to collect a subscription fee.
Show more
Don't bother
If you want a harness for your local inference build this is not the app for you. $50/yr price tag to be able to use your custom endpoint. I understand the dev needs to make money somehow but it completely defeats the purpose of blowing money on GPUs for free local inference.
Excellent
Once optimised, provides a vastly superior LLM interaction experience.
This coming from an individual with very little knowledge on the subject, still, very impressed.
Good work gentlemen.
This coming from an individual with very little knowledge on the subject, still, very impressed.
Good work gentlemen.
Looks polished but unusable
It as if the owner of this app has never used it themselves. The font is too small, you’re unable to read the LLM response at the still generating. Meaning, if your response is four pages long, wouldn’t it be nice to be able to read the top while it’s still generating? You can’t do that here it just keeps blocking you… Many other defects that are surprising for something that requires a payment when there are apps like Cu that far exceed these capabilities for free.
Show more
