{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreid35mlxdufvbnigguc77ogplvkwrnykarfdszpklqelilofy5xnbm",
    "uri": "at://did:plc:lk3jfj3zq4k4wxnk474axylu/app.bsky.feed.post/3mo7t6n3chen2"
  },
  "path": "/t/ive-been-trying-to-contact-the-developers-to-suggest-several-fixes-for-severe-issues/1383643#post_1",
  "publishedAt": "2026-06-14T02:21:00.000Z",
  "site": "https://community.openai.com",
  "textContent": "B\"H\n\nI know I’m not a developer, but I wanted to suggest technical improvements for the developers.\n\nAs we all know the website and app get severely lagged and crash, even with very high end devices\n\nWhat ppl don’t know, is this issue is amplified EXPONENTIALLY when it comes to custom GPTs with large response payloads.\n\nMy own custom GPT is essentially an agent that can talk to the client’s own computer, bulk read files and execute commands etc . Many kilobytes or even megabytes of data is returned TO the chatgpt chat, and it severally lags both the app and the website\n\nOften times I have to force close the app several times between messages, even in a high end android (pixel 6 pro). My own GPT just uncovers the underlying issue, it also exists in general\n\nHere’s a message I sent you the support team when the hope that it would reach the developers, to attempt to fix that:\n\nB\"H.\n\nGood evening!\n\nI’m the author of the custom GPT called “Awtsmoos Shliach (Agent)”\n\nWhich makes a lot of command see s sent to my server to get it to work\n\nIt does work fine, but the issue is the actual chatgpt app and website itself\n\nNot only is it an issue with my custom GPT app, but it’s highlighted more than others\n\nWhen a chat gets even relatively long in general, even though I have a really good computer, the entire website often crashes and gets super choppy. This is amplified when a custom GPT makes lots of action calls especially on my pixel 6 pro android phone, very good service, the app literally freezes after a few messages.\n\nI am a programmer myself so I can tell the main issue is that the app and website keeps all of the messages AND all of the data sent and received from custom GPT calls IN the DOM itself as many complex nested objects\n\nThe issue would im easily be fixed by virtualized scrolling, just prune out the older messages every few messages, and fit API calls and custom actions, do not add them to DOM at all, only if the user manually expands the thinking area then it can dynamically add very few parts to the Dom, then add they expand each action it can add more, then immediately remove if they collapse\n\nI have several prototypes of this for my own ai all with other ai providers, but I just wanted to suggest this to you, because your app is literally crashing every few messages even on a very high end device (using my custom GPT at least, which has large results from actions, but even in general chats if the chat gets too big)\n\nIt probably hasn’t been tested as just with custom GPT apps that have large payload responses. But it would very easily be solved by simply not adding any of the response data to DOM unless the use explicitly expands it\n\nI know you guys had a big team of coders and stuff, I just really wanted the app and website to work\n\nAlso the website, as mentioned, is a very big issue even for basic chats, even for computers with very high end RAM. It’s just inherently broken due to holding all elements in the DOM\n\nIf you want me to help I can offer to help code it for you for free, but I’m guessing you probably have specific ppl to do it\n\nJust thought I should reach out, so the app and website can fully work",
  "title": "I've been trying to contact the developers to suggest several fixes for severe issues"
}