Working on the Zapstore app, using Claude Opus 4.1 MAX (on Cursor)
MAX retardation. Is there a chance Cursor is rugging me and providing lower quality LLMs?
My prompts and context are more than good enough.
Discussion
Anthropic itself probably keeps tweaking their models. I have noticed the performance vary widely while using Codebuff, which uses Claude models internally
I’ve only used Opus when it was free credits, but it doesn’t seem worth 10x more
Use GPT-5. It's stable and good for every task basically.
I've noticed a roll off in quality lately. Very explicit instructions on how to construct a string lol