Used the Weweb AI to build a few incredible things. The only problem is that for very little usage, i am burning through tokens like nobody’s business - i think 25M tokens just today which is all the tokens I have. I wouldn’t say i used it a lot.
My suggestion: After the AI is done, ask the user if they want to keep the output. if they dont or didn’t say they want to keep it, then don’t charge them for it.
This is similar to how Intercom AI or other AIs work, pay only for resolved conversations and not for everything even when the AI is malfunctioning…
Because I am running into this problem whereby sometimes it shows as “success” but then the spinny wheel thing keeps showing the the changes arent even shown but the tokens are still running. i have no choice but to refresh the page and have to try like 4 - 5 times to get the job done.
Also, many times it says it’s done but then i don’t see the new thing being added. So i have to refresh and redo again, which also costs tokens.
So now it looks like im out of credits again despite upgrading 2x T.T
these are all fair points! We are working on that spinning wheel problem and on the AI saying it did the thing without actually doing it.
Should be fixed by next week. Can you send me your user email in DM? I’ll send you free tokens so you can keep working in the meantime.
This is a real problem! I have spent a lot of money, wasted on it not actually producing anything! This really needs to be fixed as a mater of priority as i feel almost like its a money making exercise at this point. Works fine for a couple of actions and then just stops, says it has created something but it hasnt.
I am having this same issue. I am burning through tokens, but I’m constantly getting errors. I am getting so many errors, that it’s making it unusable. I went through something like 10M tokens yesterday. Shockingly, around 2M of them was it just spinning and spinning for a good 20 minutes just to do nothing. Something is broken.
This is a really great tool, and has so much potential., but the burn-rate for nothing is beginning to make me deeply frustrated for what I’m actually paying for.
Burning through 25M tokens with minimal use—especially when results don’t even load properly—is super frustrating. It would be much fairer if tokens were only counted when we actually keep the output. Kind of like how سورة الانفطار مكتوبة reminds us that every action is recorded with intention and fairness—nothing wasted without purpose. Hoping the team takes feedback like this seriously.
Thanks everyone for sharing your thoughts here. We’re currently working on improved context management to help optimize token usage, with a release planned for July. We’re also in ongoing discussions with AWS and Anthropic to reduce token costs and consumption - benefits that should roll out to all users in Q4.
