Android App - Can't enable beta features with Plus account subscription.

[soft compass]

Bug Report: 2024-02-25

Hi - None of the standard bug report fields are applicable until the "Additional information" section.

Steps to reproduce:

1. Sample text
2. Sample text

Expected result:

Sample text

Actual result:

Sample text

Additional information

Browser: N/A
OS: Android 14 with Latest security update (Feb 5th, 2024) on Pixel 6 Pro.

Hi, I am a chatGPT plus subscriber. However, when I attempt to use the plugin option of GPT4 in the Android app I am unable to do so.

I have significant visual impairment so it may simply be that I'm unable to locate the beta features slider that ought to enable the plugins model of GPT4. Is it hidden somewhere that I'm unable to locate, or is it simply unavailable?

When I click on my name at the top of the left hand panel then click subscription (which does show that I am a plus user), a menu/page opens to the right showing a list of the Plus features.

At the bottom of this page is a cryptic message indicating that I don't have permission to change my subscription as I signed up for the Plus subscription through a different platform Screen shot included.

I am running the latest version of the Android ChatGPT app available from the Google Play store - screen shot included.

I've attached two screenshots. One shows the extended menu with the cryptic message at the bottom, the other shows the left hand panel providing the ChatGPT version number.

I'd appreciate any help you can offer as soon as possible in order to resolve this issue.

Thanks and regards

[winter beacon]

Thank you for reporting a bug related to ChatGPT.

If relevant and not already provided, we recommend adding a link to the associated ChatGPT conversation and uploading any relevant images here for easier troubleshooting.

➜ Need to contact OpenAI support? Visit https://help.openai.com

[whole loom]

Hey there, the Plugins model was never built in to the app versions because plugins have been supplanted by GPTs. It's possible and maybe even likely that whichever plugin(s) you were hoping to use have a GPT correlate available in the GPT Store. In the chat history sidebar, there's an option to "Explore GPTs" -- this will open the Store and you can search from there!

However, if you still want to use the Plugins model, it is still available on the web version at chat.openai.com, mobile and desktop versions. In a browser, you can navigate to https://chat.openai.com/#settings/BetaFeatures and enable the Plugins model there. Then, it will be available in the model dropdown selector in a new chat (still web version only).

I believe you do have the option to start a Plugins chat on the web version and then to continue it via the app, but the UI might get a little funky since plugins weren't designed to be included in the app. But, the functionality should be the same.

Let me know if you've got any questions!

[soft compass]

Hey @whole loom - Thank you so much for shedding light on the issue! That subscription message really threw me off. It was driving me nuts searching for something that didn't exist - par for the course these days unfortunately, lol.

Incidentally, I received a notice the other day about deprecating the Plugin feature / store. I believe the store will remain open until March 8th, then completely discontinued on April 9th-ish. I believe they addressed the behavior but I don't recall the specifics. I'm reasonably sure the conversations will no longer be advanceable after April 9th, but unsure about the month in limbo.

That notice also described a new Beta 'Mention' feature for incorporating @CustomGPTs (probably in prompts, though possibly via Custom Behavior). Do you happen to know whether this feature allows the inclusion of multiple CustomGPTs?

Unfortunately, deprecating the Plugins and replacing them with GPTs leaves a gap in that multiple plugins could be inclided, while we're currently limited to a single Custom GPT. Allowing multiple @CustomGPT mentions would help close this gap.

This brings me to my final point. Do you know whether Beta features are planned for the mobile apps? If not, the same issue will arise once the @CustomGPT mentions and any other Beta features are added.

Is there a formal public product road map available so we don't waste support resource time answering this sort of question?

Thanks,
Chuck

[winter beacon]

Is this issue solved?

If you believe this issue is now ready to be closed, please press the button below.

[whole loom]

Hello again, sorry for my delay!

As for the @mention feature: you currently cannot @mention multiple GPTs in a single request, but you can use @mention for multiple GPTs over multiple messages in a single chat.

I think I know what you mean: with Plugins, you can enable three plugins and then ask for all three of them to do something in a single request, and that indeed doesn't seem possible with the current @mention feature.

However, what you can do is @mention an indefinite number of GPTs in a single chat -- beyond the limit of three with the Plugins model, so a bit of a trade-off there!

As for bringing features to the app: I wouldn't expect that Plugins ever get added to the apps as they head toward deprecation, but I would expect the @mention feature to make its way over to the apps eventually, but this is just my guess! They've certainly done things to try to create parity between the web and app versions, and this one seems like a great candidate for porting to the app, so I'd bet on it, personally. (In the meantime, you can of course use the mobile version of ChatGPT for full @mention abilities! Again -- can @mention any GPTs, any time, as long as you're using GPT-4!)

I don't think there's a formal roadmap out there, no -- we get surprised on the regular here 😉 But no worries on wasting support resources -- please always ask whatever questions you have, most of us here are community members like you just helpin each other out! OpenAI team members in this server are indicated with yellow usernames and also have the logo in their name, though I know you've indicated you have a visual impairment so please let me know if you have any more questions about this or anything we've chatted about!

[soft compass]

Thanks for the detailed answer - very helpful!

That's all for this one, but I do have a fairly lengthy question about how best to use files in the #community-p/custom-gpts (discussion tag) channel wrt Best practices for uploaded data files when building custom GPTs through the app/web (not via API). I could copy it over here, but probably doesn't qualify as an actual bug (other than possibly lack of documentation?) and would prefer not to cross-post unless you think it makes sense. Any chance you could take a look and either point me towards documentation /other helpful resources or provide guidance?

And yeah (I mean no😁) I didn't expect an umm public Roadmap.

[winter beacon]

Is this issue solved?

If you believe this issue is now ready to be closed, please press the button below.

[whole loom]

Good question! So are you asking about attaching knowledge files when you're building a custom GPT? If so, first of all: similar to the @mention feature, building GPTs is not currently possible in the app, but you can do it on both the desktop and mobile versions of the web version.

<chat.openai.com/gpts/editor> is the URL to get started with that. There are three main sections of the GPT builder: Create, Configure, and Preview.

You can think of Create as a chat interface that fills out the Configure section for you. This means that you also have the option to skip Create and just go straight for Configure directly, which is how I usually build GPTs for myself.

Configure has a few sections itself: Instructions, Knowledge, Abilities, Actions. Instructions has an 8000 character limit and lets you instruct the GPT on how to behave. Knowledge lets you upload up to twenty 512MB files. These are like reference documents that the GPT can search through for you, and return relevant information from. Here is a helpful help article on knowledge files: https://help.openai.com/en/articles/8555545-file-uploads-faq

Let me know if you've got any more questions, sorry if you already knew some of this or if I'm misunderstanding! #community-help and #gpt-models are good potential places for further questions too, just so you know! Lots of helpful folks in this server all over.

[soft compass]

Thank you, very helpful.

First - Yes, knowledge files for building a custom GPT.

All good wrt app vs web. I bounce around a lot and just got caught by surprise with the lack of beta support on the Android app.

I checked out the article and quite a few others, but still don't have a great sense of either how to select and/or format files to upload for best impact (detailed description in the other post). For example, if I upload a dozen PDFs of textbooks GPT4 had already been trained on, would that still improve the custom GPT's responses in the desired domain? Similarly, would the type of 'training data'files described in the other post help accomplish that goal and if so, what is/are the best format(s) for these?

Also that article referred to the file upload as a 'capability', but I didn't get the sense that term was used in the same context you used it.

Is there any other more thorough documentation that includes examples or best practices? I don't mind reading API documentation if that would help shed some light.