Did it actually used 23 requests, if so, why?
If this is a measurement for something else, what does the 23.8 stands for?
1 Like
Claude 4 Opus is a MAX model, that means it charges by amount of input, thinking and output tokens from API usage. Tokens are parts of your request (instructions, rules, files attached, fetched or processed) and also the output of AI (text & code).
For example if the usage of Claude 4 Opus in Max mode resulted in e.g. 23.8 requests charged = 23.8 x 0.04 (USD/Request) = 0.952 USD total
This has changed with Cursor 0.50 from a cost per request and cost per tool call to a per token used pricing. It avoids charging for each edit or file read tool call as 200 such calls are included but you have to manage the context (tokens) by yourself.
Such Max models are recommended for those with experience in prompting advanced models and managing tokens. You can use the regular request priced models in non-Max mode with fixed per request cost.
Cursor team has mentioned this also on the Changelog pages for app updates.
Additionally the pricing page shows info about requests and Max mode
Pricing Update for Claude 4 models by Cursor team:
At least you are able to use it. Every time I have tried to use it I get that is has too many request and have not been able to use it once since its been out. Sonnet as of now is my best friend.
1 Like