CursorAI "unlimited" plan rug pull: Cursor AI silently changed their "unlimited" Pro plan to severely rate-limited without notice, locking users out after 3-7 requests
CursorAI "unlimited" plan rug pull: Cursor AI silently changed their "unlimited" Pro plan to severely rate-limited without notice, locking users out after 3-7 requests
Nowadays, anything "unlimited" is highly likely to be fake. There's just too much potential for abuse and nobody is going to finance that.
Cursor is also US, so their terms and conditions can change at any moment without following any rules whatsoever if you're in the US. Probably for EU customers they might have a problem, because they require notification of changes and to give the user the option to reject the changes and cancel with no supplementary charge.
Everyone is guessing how much better llm can get, I wanna know how much it really costs.
I assume all ai companies are operating at a loss, but to what degree? Is all the RAG stuff a fad or in the future a significant part of api requests will go through llm? Can you stuff million lines of code into the black box and expect it to come up with a reasonable answer at a reasonable price?
I can't answer any of that without knowing the algorithmic cost of all this stuff, and whenever I try looking the answer ranges from "well its a few times heavier than a google search" to "a single request can power a small household for 6 hours"