>Claude usage limit reached. Your limit will reset at 5pm (America/New_York).
>/upgrade to increase your usage limit.

>>105593855 (OP)
>didnt get a second phone number
about on par with what i expect from a chatbot programmer

>>105593863
I paid for Claude Pro and even then the limits are ridiculous.

>>105593855 (OP)
>>105593863
(cont.)
ur from the us, right?
cant you get a prepaid phone for 20 bucks at walmart?
lel, even in jewrope its not a problem to get another phonenumber
its gonna cost you somewhere around 10 bucks, just you have to give your id

>>105593875
yeah no thats a joke
greed overload

>>105593875
(cont.)
i cant imagine how jarring it must be to have your tool suddenly tell you it will take a break during a state of flow

>>105593891
I think they jewed the free version because after one prompt it runs out of tokens. That's the reason probably, they want you to upgrade. Which I did but the limits are still abysmal. The longer context window is great, but I wanted to try out Claude Code their CLI tool that uses grep and edits your project files in place and it requires Pro. Actually it's not even that great and aside from the more tokens in the web interface I'll probably go back to using the free version and programming things manually because waiting on Claude to finish wastes a lot of time to then end up giving me not working broken code.

>>105593935
i know it reprompts itself and thats how it may run out of tokens so fast/be so inefficient with tokens
but the client doesnt fucking care
and your experience is the result.

i think the paid tier is another tier which goal is actually advertisement-
this one aimed at execs and such
i bet theres another "extra pro" tier with no limits, or deals for companies

and the broken code is kinda unavoidable given how an LLM works
especially when your code allows for side effects
the code is subdivided into contexts, but side effects are cross context relations and LLMs just are bad at that

>>105594027
I know all these things and that it just injects the system prompt and your "settings" as a prompt and all that other bull like <think> bpe token. You hit the nail on the head though with the user experience, it's horrible. Yup I don't think they even care about selling "Pro" because they're not breaking even any time soon.

>>105594092
>they even care about selling "Pro" because they're not breaking even any time soon.
i heard that one too although i find it hard to believe to be honest
i didnt get into gen ai yet but i have some experience with ML in the context of pattern recognition
and while training a model is quite compute intensive, running that model certainly is not
i use a gpu to train, and it takes days but then i can run my models in a single thread, on a cpu
i think the relation stays the same, only scales up with gen ai

but even then
how expesive 60 seconds runtime of a gpu can be?

just run locally

>>105594201
they aint gonna give you their latest model
or theyre profoundly retarded. which they obviously arent since they still managed to get the thing to (somewhat) work

>>105594235
skill issue

>>105594253
retard

>>105594263
says the nigger who cant run models locally

why do so many people ignore gemini via aitstudio, have you guys used it much or no?
>1mil context window
>they block your requests if you go crazy sending like 5 in a minute
>there is no daily limit bullshit
>less censored than claude
even if the code is "worse" than claudes, just being able to constantly use this all day for free with your entire codebase is insane. people sleep on it i dont get it, it's actually really good in my experience.

>>105594391
>delusions, my only friend
retard

>>105593855 (OP)
Imagine still being a subscriptionpoor. Just use the API.

>>105594535
post nose

>>105594502
at that point you're better off just using free ChatGPT or Copilot, it's the same "worse" code