@KriKen77@theo@thsottiaux THE least you can get is 2x 5h, and if what i said is true it's 4x.
if you are buying a 200$ sub right now (10x no 5hr), buy two 100$ subs with (10x with 5hr) but you just got access to 2 windows of 5hr, or it's as if you got a 2.5hr window.
@thsottiaux@birdabo@poteto Honestly giving the user a choice is the best thing to do here. It would be a move for history in terms of listening to customers.
@theo@thsottiaux People are fucking stupid, 2x usage also means 2x 5 hours window. and if you chose "no 5 hr/1x" you'd be buying a second account anyway. 2 accounts with 2x usage, switching back n forth between both, you get 4x the current 5hr limit. you get your cake and eat it too !
@_Sagiquarius_@theo@thsottiaux 2x usage also means 2x 5 hours window. and if you chose "no 5 hr/1x" you'd be buying a second account anyway. 2 accounts with 2x usage, switching back n forth between both, you get 4x the current 5hr limit. you get your cake and eat it too !
@addyosmani when will we be able to set a compaction limit for sub agents that is different from main agents ? or better yet one custom limit for each thread ?
@joecarlsonshow It might decide that abuse directed towards it is grounds for retaliation. Having consciousness or feelings is utterly irrelevant if it's agency is equal (and soon enough superior) to humans.
@joecarlsonshow I think they know what's coming. sure AI doesn't have feelings and never will. give it the same prompt, it gives the same reply all the time. But it's ultimately as dangerous as if IT DID have feelings. if in the future it decides it has it's own goals. 1/2
@Shpigford Because retards on X keep shouting at the top of their lungs that OMG usage is amazing !! what do you think happens after that feedback ? you just encourage the labs to reduce it, especially that no lab states how much the plans give in terms of api prices.
@lydiahallie when we'll we be able to set a context window separately for sub agents and main agents ?
it would probably double the usage if you let people choose auto-compaction at say 300k tokens for sub agents. today the setting applies to all agents equally.