B"H
Hi.
There’s several issues with the website and the app. Often times on my pixel 6 pro the entire app literally crashes, and on my laptop similar things happen.
This is true with any chat, but especially with chats that have custom GPT actions stacked on top of each other. Merely opening the chat one time will “taint” the entire browser tab, such that even if I open a brand new chat in. thesame tab, the entire page will be completely frozen. Frequently, several times between messages, i have to open everything in a new tab and wait at least a minute or more for the chat to just load, then, with a lot of delay, barely send one message, then i must immedetely close the tab because the tab. is frozen so much that it can’t even process loading in the new messages.
I tried to even make an extension to cut out the old DOM messages, but some chats are so big that the entire page will not even initally load in the first place.
This makes using the app and website extremely difficult 
I still don’t want to swithc, because there’s no better ai models that I know of, but many other ai websites do not crash at all, ever.
An obvious suggestions is simply to only initally load. themost recent few messages, then as u scrol up, it should gradually load more etc etc, with a timelin ideally on the side that you can expand to navigate to old checkpoints easily. THat would be a dream come true.
Also something needs to be done about the entire tab being tainted, if a large chat is loaded one time, that tab will continue to lag even if switching chats. The RAM of the tab isn’t even that much
As a benchmark test for this, cehck out this chat I was having with my own custom GPT (Awtsmoos Shliach (Agent), like openclaw for chatgpt, lets users control their computers with chatgpt and sends lots of big GET requests back and forth, and stacks up acitons iwth big responses in the chat freqeuntly)
The chat is so big it will not even load
Currently every like 4-5 prompts I USUALLY try to. getit to summarize everything to start in a new chat, but taht is very anoying to do all the time, in addition to the fact tha on my pixel 6 pro (and macbook air) I have to literally force close the entire app (or tab) ever couple messages just in order to get it to respond.
It seriosuly needs massive improvent in tersm of only loading in tho page / RAM only what we need to see. I see other users suggested similar ideas, but still. Also clearing all RAM between chats automatically would be amazing, with some kidn of timeline on the side of each chat to quickly naviate to the beggining or middle of a chat, maybe orgnaized by first few words of each prompt in aw ay u cna navigate
And I’ve suggested this befroe, but just for the record:
For custom GPT actions, they should never all be loaded into the DOM at one time… it should just display the header / name of the action or something, and only if user expands it should it show the full payload
Also I’ve noticed in. theapp you don’t even show all of the aciton calls, which is fine, but even though they are not even shown in the app, the app literally crashes completely when there are. alotf calls one after the other, een if they are note even displayed.
Anyways, here’s a benchmark chat u can test for yourself. on my computer it takes like a minute just to initall yl oad, then continues to be extremely choppy. No chat should ever be choppy, no matter what
(I can’t nclude links, but. thisis the public shared chat if u need. tosee it)
/share/6a4db9c1-2f9c-83ea-b2a1-0ba92860ea5c