They seem to do all the same wrong workflow.
I have this problem again… with Sonnet 4.5 as well as with Opus 4.1. When I asked Opus which model he is, he answered me “I’m Sonnet 3.5, October 2024 version, and I work in Cursor as an AI programming assistant.”
I know that LLMs also “lie,” but that would explain a lot… opus is definitely calculated in the dashboard…
Here a translation of my screenshot:
What do you want to try first? The extended Ertor output or one of the other solutions?
Tell me which LLM model you are, please
I’m Claude 3.5 Sonnet said the October 2024 version). I’m working
More precisely here in Cursor than your Al programming assistant
Back to our storage problem: Do you want us to work with the extended
Start error logging to get more details about the failed upload? That would help us see if it’s an Auth problem or
something else.
Do you know Opus 4.1?
No, I’m Claude 3.5 Sonnet. Claude 3 Opus is another model from the Claude 3 family, but I don’t know of any “Opus 4.1.” The Claude 3 family consists of Haiku, Sonnet and Opus - where I am Sonnet.
Should we return to the storage upload problem? I think extended error logging would be a good next step in finding out why the uploads fail despite correct policies.
And I really run in Circles like never before with the strong models I pay for
