I even cannot post a link…
Hi @Marv_Yu ,
the topic has been re-opened.
Your feedback has been noted. We should look into this and thank you for raising it.
This is not a ChatGPT support forum.
The ChatGPT category will auto-close any posting made, it is a historic category.
If you try to post there, you will be strongly dissuaded with warnings:
The topic title appears to be about the ChatGPT user interface - and ChatGPT’s UI is one of the topics with consequences if opened.
Personal opinion:
With Codex and ChatGPT now combined in a single app, it is becoming increasingly difficult to sort and filter hundreds of posts while distinguishing between ChatGPT-related and developer-related topics.
The clear separation made much more sense in 2024 and 2025, but this may need to evolve as the products become more closely integrated.
I would be glad to hear some additional perspectives on this.
Is it a developer:
- consuming a subscription product where the community is powerless to help, because they have complaints about the service quality that will reach no audience of assigned personnel, help.openai.com is the support channel, and there will not be discussion of any programming;
- developing an AI application with OpenAI API endpoints and backend products and needing support for how to use them, or needing bugs corrected in the API products in order that they operate per specification?
There is a clear distinction that remains between:
- “I want to prompt AI to make me anime-themed tetris” but I don’t like the app’s user interface;
…versus…
- “where’s the documentation for the backend resize algorithm of sending images to gpt-image-2 on the edits API, and how are patches billed as tokens by input dimensions that result?”
Yes, those are the easy examples.
What about:
- I need better project and chat organization.
- Live voice is not spawning the subagents I need for the given tasks.
- Widgets do not render in Codex Desktop.
- I am running into drift during larger Codex tasks.
And there is a even larger grey zone.
Developers and builders are increasingly using the same surfaces as consumers for their work.
How about:
Can your peers help with any of that?
Would issue trends and feedback be better surfaced in aggregate by help.openai messages instead arriving here for nobody to read except “me too” posters?
Is it consumer-facing subscription plan optional software product, and the issue is with software quality and not API services?
Is it a model behavior or quality for which OpenAI only takes feedback in a thumbs-up form?
I can fix all those concerns in my own software or in improving other developer’s use, but not OpenAI’s.
And: what kind of support per-user is expected for $20/mo … vs a $200,000 API cap per month before an account rep asks if you want a service commitment contract.
I seeeeeeeee. Thx for your mention. :)))))))))))))))))))))
However, I truly wanna discuss something about the recent desktop version which combined “codex and chatgpt”. Maybe I should use “codex” category?~
Exactly this. The first thing that comes to mind is image generation, for example comparing moderation, prompting and transparent-background in API and ChatGPT. These are overlaps we have discussed in many different threads.
I would also add Work, which in practice sits somewhere between Chat mode and Codex. I personally use Work a lot and think it was a great addition.
So yes, in my opinion, the ChatGPT category could use another look.
While I can still access Codex locally through a terminal or a shell, the state of the app, as VeitB pointed out, forces an evaluation of policy.
The branding in the web access and the downloadable app have replaced all Codex branding markers, with ChatGPT.
People are more likely to ask developer questions without realizing it because of the merge, and even pinned chatgpt sessions migrate over to the work side as demonstrated in the above screenshot.
While I’m fairly certain we’re long past the point where this community will get swarmed with ‘ChatGPT-5 IS ALIVE’ threads, I think the issue to evaluate is the types of inquiries or complaints the forum would endure are around the novice developer area.
Meaning people are likely going to want developer-related questions answered without fully realizing themselves that the question is development-related.
While that can certainly be annoying for some here, I feel like recognition that these software novices are tomorrow’s developers is a good stance to consider.
People generally won’t understand how to avoid the need to refactor, because they don’t understand software architecture.
The transition from vibe coder to software development might be a very good place to center the policy around.
Today’s vibe coding lapins, are tomorrow’s developers.
I agree with parts of your saying. Even old ChatGPT/Codex user will be confused by the recent desktop features. Be honest, I like the merge of these two product.
However, the problem is the merged product seems not be completed…:(((
Yes, that’s why it’s constantly being developed, my friend.
Thing here is, that there are overlapping domains, like developing a plugin or a skill for ChatGPT is part ChatGPT part Developer. So it’s not black or white.
With that being said, I do see most of the ChatGPT threads are oriented to the ChatGPT experience rather than that what I mentioned.
There is one thing I did notice between Devs and non-Dev support on Discourse, and this is probably a defining factor: The presence of the support team on Discourse when it comes to ChatGPT as a product is noticeable. On Discord, which is the intended non-Dev channel, this doesn’t reflect the same way.
This brings it to, if people see the presence of Support in one or the other platform, then they go where they think they can be heard.
Even if both have means and channeled ways to get the user’s matters.
As to how to solve this… that’s for Support to handle it. Not easy when products overlap in domains.