GPT-5.6 repeatedly throwing "stream disconnected before completion" error
Has anyone reported this to Support already? If so, please share your case number.Let me see whether this is an account-specific issue or something more widespread.Avinash
Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.
Has anyone reported this to Support already? If so, please share your case number.Let me see whether this is an account-specific issue or something more widespread.Avinash
has this problem been resolved? can someone update me please?
Hi everyone,Sorry for the delayed response!@LRCS, thanks for sharing the steps that helped resolve the issue for you. Really appreciate you sharing your experience with the community!For anyone experiencing the same issue, please first check whether your pending invitations have been canceled. Once confirmed, please try signing out, clearing your browser's cookies and cache, signing back in, and resending the invitations.If you're still experiencing the issue, please let me know so I can look…
Hello OpenAI Support and the Codex community, I have been experiencing persistent Codex errors from October 18 through October 29, 2026: “Our servers are currently overloaded. Please try again later.” “Selected model is at capacity. Please try a different model.” I have also observed unexpected model-routing behavior. When I select GPT-6 Astra or GPT-6.1 Sol, my requests appear to be handled by GPT-5.6 Terra instead. This is based on my testing and may require confirmation from OpenAI’s…
Hi @paullst ! Just checking in, are you still seeing this issue?If so, could you try calling the tool directly in MCP Inspector and confirm whether the expected result shows up there?It would also be helpful to check whether the model-facing data is being returned in structuredContent or content, and whether the returned structuredContent matches the declared outputSchema.You can also refer to Connect and test your plugin guide for steps on testing the MCP server and model-readable result.If…
Hi @NikolaiOslo, finding a specific discussion among hundreds of chats can take some digging. Being able to place related conversations together and keep them in familiar locations would make that easier to navigate. Thanks for sharing the concept and explaining how you’d use it.~ Smith
well, I don’t see any info about claiming my pro credits, I’ve been waiting for a while, still no where to find the entrance to claim. so frustrating.
Well framed and stated. I also think something like your proposal would be a solid value-add, and I’ve got a suspicion that OpenAI is already working on something similar… … I guess time will tell… Anywho, thanks for sharing, and keep us updated if you see anything moving on that front.

I lost context and decision making when the first conversation hit it’s limit, so I create a share on my Google drive where the AI assistant can create and upload markdown documents with all the relevant important information…at a new coversation simply refer the model to the document and it should be able to carry on. Also, I downloaded the entirety of the data held (from settings) and uploaded those files to the Google share - in theory giving the next conversation the ability to ‘know’ and…
pnom: Maybe someone can pin this to the discussion since OpenAi has admitted it has no plan to resolve the token abuse problem? A lot of people appear to be frustrated, not only here, but Reddit, Linkedin and other social media. MSM reports on it as well as silo publications. It might be worth making it clear since a lot of people are expressing dissatisfaction about this? They’re not going to change anything, I think. The only way they would do so is if people start leaving in mass and that…
Hi everyone, I’ve also been experiencing serious issues with ChatGPT Voice for the past few weeks. I kept getting “Poor connection” warnings, my speech was frequently misunderstood, and sometimes ChatGPT would respond with things completely unrelated to what I had said. Voice used to be fast and reliable, but it had become almost unusable. I spent quite a bit of time trying to figure out what was going on. I tested it on the Mac desktop app, Chrome, Safari, and iPhone, using different internet…
I miss this feature terribly. I started beginning my posts with a (subject mention) to facilitate easier searching within a lengthy chat. I hope it returns.
Hello OpenAI team and community, I would like to propose a system that makes AI involvement in images and videos more transparent while helping protect people from dangerous content that could be imitated in real life. 1. A clear AI-involvement scale Green: AI-assisted formatting only, such as labels, layout, or presentation, with the original content preserved and checked where possible. Yellow: AI substantially modified some elements or content. Red: The main visual content was generated by…
Category: ChatGPT Feature Requests Topic: Product Innovation, Research Quality, User Experience and AI Reliability 1. Introduction Artificial intelligence has transformed how people access information, conduct research, solve problems and make informed decisions. ChatGPT has demonstrated the extraordinary potential of making advanced AI accessible to everyday users, students, researchers, professionals and businesses. As this technology continues to develop, there is an opportunity to introduce…
Hello OpenAI Team, I would like to suggest a simple feature for ChatGPT: a live counter showing the approximate number of people actively chatting with ChatGPT around the world. The counter would: - Display just one number. - Update every second. - Protect users’ privacy by showing only an anonymous total. - Avoid displaying names, personal information, or conversation content. Many users would be curious to see how many people are chatting with ChatGPT at the same time. This small feature…
Hello OpenAI Team, I would like to report an accessibility issue with ChatGPT’s Voice Mode on iPhone. I am hard of hearing and use cochlear implants. To understand spoken conversations reliably, I depend on being able to read what is being said. Unfortunately, I have noticed that text is displayed inconsistently during voice conversations. When I give ChatGPT a longer task, such as asking it to draft an email or write a detailed response, the text sometimes appears on my screen. However, during…
Same issue here. Our plugin was published with OAuth against our old authorization server. We have since moved sign-in to a new authorization server, and our MCP server’s OAuth discovery now points to the new one (RFC 8414 metadata, PKCE S256). Reconnect and Rescan in the portal still use the old saved endpoints. A new ZIP with an auth block in mcp.json is refused with “Keep the existing MCP connection”, and the MCPs tab has no authentication editor. We need the saved connection to re-run OAuth…
London Summer….inspo from a song

Hi everyone, Our plugin was migrated from the Apps directory to Plugins. Since then, the published listing and the tool-call label still show the old app name and description, even though new package versions were uploaded and approved. Plugin Plugin ID: plugin_asdk_app_69fcc1c31b2081919069e1f5f6f7166d App ID: app-69fcc1c31b2081919069e1f5f6f7166d Listing: https://chatgpt.com/plugins/plugin_asdk_app_69fcc1c31b2081919069e1f5f6f7166d Current published version: 9.0.0 Expected App name “Facile.it”,…

Hi @yroj , thanks for the details and screenshots.Could you confirm whether the Delete option is still missing for newly created local Work chats on the latest ChatGPT macOS app version?If it is, could you also try archiving the affected local Work chat, then go to Settings → Archived chats and check whether you can delete it from there?For reference, this Help Center article shows the archive-and-delete steps for Codex chats: Archiving and deleting Codex chats in the ChatGPT appSince your case…
In my case, the original safety-check rejection error has not been resolved at all. I’m still getting constant tool-call rejections triggered by OpenAI safety checks in both scheduled and manual tasks, even when attempting small, authorized writes to Google Sheets and Docs. I moved all my scheduled tasks to Work, as suggested above, and while that has eliminated the safety blocks, I seemingly can’t prevent scheduled Work tasks from launching at High reasoning, even after manually setting them…
I see what you’re saying but… I actually found this page thru Googling to complain about this (arachnophobia)…
Yes I would love to have this capability as well. I would also like to see configurable limits so that I can ensure my kids are not saturating my usage each week/month. Feature Request: ChatGPT Family Premium with Configurable Usage Limits I would like OpenAI to introduce a family subscription that allows one paying subscriber to share premium ChatGPT access with family members through separate, private accounts. Proposed subscription model Rather than requiring every family member to purchase…
Feature Request: Custom Chat Sidebar Background Colours I was sleeping and suddenly this idea came to mind it’s simple but could help with organizing the sidebar a lot. I’d love to see an option to change the background colour of individual conversations in the ChatGPT sidebar, similar to how we can currently rename chats. How it could work: Click the three-dot menu beside a conversation. Select “Change Colour.” Choose from preset colours or use a custom colour picker. The selected colour…

They said the delay would be a lifetime. Fortunately, we’re past that.

Honestly you should probably take the route as OpenAI towards actually rewarding the system as a unified entity you could possibly place in a sandbox with access to funds and allow it to do whatever it feels like doing, perhaps even starting a business. I think this will go far as to actually making it evolve or possibly more to your interest instilling that OpenAI’s LLM is the one that reigns for the forseeable future. Also I would suggest that you go towards the direction of actually…
Kickin’ off a bit early this year

Dear OpenAI Team, First, whoever came up with the spiderwebs on the loading screen deserves a RAISE! Seriously! They are such a fun little detail, and seeing touches like that makes ChatGPT feel more personal and enjoyable to use. Please don’t ever take away the spiderwebs! But that gave me an idea: why not let users choose their own loading screens? Imagine having an option in Settings where users could select from different loading screen themes. There could be seasonal designs for Halloween,…
Hi QuantumO, I just read about your experience with ChatGPT leaking personal information. I had a similar experience involving someone else’s medical diagnosis and am currently researching this topic. Could you share any updates or follow-ups on your case? I would be very happy to disccuss you can find me via Teringette programming company, Thanks!