ChatGPT is a confusing mess right now. It seems like a few months ago they embraced a new product strategy (which is good), but there's still lots of legacy features like GPTs that feel really out of place with newer releases like projects and tasks.
What bothers me the most is that even for power users, it's extremely difficult to know which tool currently works with which model on which platform (web, mobile, desktop). The feature matrix is incredibly complex.
In the normal chat, I can connect Google Drive, but this doesn't work as a data source for GPTs or projects. Advanced voice mode can access custom instructions and memory, but doesn't work with projects or GPTs. o1 can now handle file uploads too, but only images. When I upload a PDF to the chat, only the text contents are analyzed, but if I upload screenshots of the PDF pages as images, these can be analyzed. Projects can only use GPT-4o as a model. I could list at least five more things that are similarly annoying in daily use.
I'm an absolute power user, and even I sometimes struggle to keep track of everything.
I wish the product team at OpenAI would focus more on removing all these complexities from the product.
@kevinweil @nickaturley @npew @embirico @TheRealAdamG @sulmanrc
23
12
245
21,034
I love perplexity and use it everyday but that statement is not true. OpenAI‘s deep research sometimes gives me much more detailed reports
7
221
19,339
it's hard to believe how much the us has been shooting itself in the foot over the past few weeks in the ai race against china.
1
1
114
11,966
honestly it's SO MUCH WORSE than what others are doing because it's not in the meeting but only recording audio output which has massive downsides. gemini notetaking know exactly who said what for example
7
1
111
15,070
"Meta should have made it clearer that “Llama-4-Maverick-03-26-Experimental” was a customized model to optimize for human preference."
this is crazy
2
1
96
6,577
How are countries supposed to cave? Many countries that have zero tariffs got hit because it’s based on the trade deficit and not actual tariffs.
3
1
72
4,694
i think it's just a mistake; they published the correct one on the launch page: openai.com/index/introducing…
8
55
23,576
the main reason many people on here talk shit about gpt-5 is because they expected a revolution compared to o3 which i think is a crazy thing to expect considering how much of a jump o3 was
3
48
3,127
wow, that's awesome; you should communicate this more/better. even on your official website it says 32k but that's only half-true then: openai.com/chatgpt/pricing/
1
44
8,025
Idk man seems like they are finally having their well deserved viral hit with Veo 3 (+ Audio)? Also the teasing before I/O from Sundar and Demis .. I think they are getting better at playing the zeitgeist. And most of the things are actually available and not just announcements!
1
39
1,575
you can still outsmart these prompt injections, this took me 2 seconds, a more sophisticated version might be needed for more complex injections but for this paper it worked
2
1
36
5,680
Nope only for 60 minutes; it’s gone again, I think this was a mistake, they said it’s gonna take a bit for Europe right?
1
34
2,462
Phew, I wish you success, but these are tough news for power users of Arc. It seems like not much will happen in the future. Very unfortunate.
30
3,061
honestly gemini's glacing is even worse than chatgpts, it's crazy. a few examples from the first lines of four CONSECUTIVE chat messages today:
This is a phenomenal find and you've absolutely hit the nail on the head.
Yes, that's an excellent idea! It's much stronger visually and in terms of content.
Yes, absolutely! That's a crucial point and an excellent addition.
Yes, that is a very good and important point.
---
pls i love gemini and 2.5 pro a lot, but the glazing and the way gemini only gives positive feedback is really bad for work stuff.
cc @OfficialLoganK @joshwoodward
1
26
6,490
i work with many agencies and marketing teams and midjourney was the undisputed king for a long time, but last year others overtook midjourney.
some examples: google with imagen 3 in image quality (try generating several people in one image with both and look at the hands), krea with the flexibility in the interface, flux with good image quality and fine-tuning possibilities …
5
27
2,135
Nahezu alles an deinem Tweet ist falsch. Die OpenAI-Integration ist lediglich für Siri und die Writing Tools. ALLE anderen KI-Features, die gestern vorgestellt wurden (dutzende) sind Apple-Modelle, die entweder lokal oder auf Apple-Servern laufen. Quellen im nächsten Tweet.
2
24
1,175
the biggest problem is that the feature would be great paired with deep research, but currently it only works with the normal web search and this is garbage for any kind of research
1
1
25
1,488
I have set up a task that should give me the most important news from the past 24 hours every morning. The problem is that I often get news that is several days or weeks old.
31
Please don't kill GPTs like Microsoft. I think 99% of the world has not yet understood the power of GPTs. I use dozens in my daily workflows. GPTs are the most important tool to be really productive with ChatGPT.
@ChatGPTapp @TheRealAdamG @willdepue @lilianweng
3
2
18
1,935
i feel like many many people on here need to stop adjusting their beliefs after every launch; it's actually crazy how much of a rollercoaster most people's beliefs are.
1
19
4,542
hmm, why is it missing the point? this is daily business for millions of agencies and marketing teams around the world, no? creating high quality coherent images and then doing some post-editing in photoshop for brand tonality is awesome, no?
3
18
670
This is great stuff. Thanks, Kevin.
My wishlist for the next update 😬
🔄 Google Docs Live-Sync: Let me attach Google Docs/Slides to projects/GPTs and have changes sync in real-time so ChatGPT always has current context
📝 Queue Prompts: While reading long responses, I get dozens of new questions. Let me send them immediately into a queue instead of waiting for the current response to finish
🚫 Don't Auto-Scroll: When I send new prompts while reading, stop auto-scrolling down. Super annoying when I'm still reading above
🏢 Workspace Company Context: Let ChatGPT workspace admins add company context that applies to all accounts - would massively improve response quality
🧠 Project-Based Memory: Account-wide memory often distorts use cases. We're a consultancy working with different clients - each project needs isolated memory, not context bleeding between clients
2
19
1,812
A wonderful example: ChatGPT Enterprise now supports reading and understanding visuals (images, graphs, diagrams, etc.) embedded in PDF files. Users can upload a PDF, and ChatGPT can interpret the text and any visual elements within that file.
Cool! But ... "and it is not currently available for GPT-based projects."
These things are so incredibly confusing in daily use for users: In my personal account ChatGPT cannot analyze images in PDFs, in my business account it works, but not when I upload the PDF as knowledge to my GPT ...
3
2
18
1,771
This is great, Kevin! It would be great if tasks worked with projects. Then I could say: Create a new Spanish exercise every morning at 09:00 and base it on our learning progress.
3
6
1,087
idk i do think there's quite a difference between short form video and attention optimized feeds and having to actively prompt/text chatgpt, wait for responses, read them even when they are personalized etc
dont you think so nathan?
2
15
367
Nope, that's wrong:
A few GPT-5 updates heading into the weekend:
- Now rolled out to 100% of Plus, Pro, Team, and Free users
- We’ve finished implementing 2x rate limits for Plus and Team for the weekend; next week we’re rolling out mini versions of GPT-5 & GPT-5 thinking to take over until your limits reset
- GPT-5 thinking and GPT-5 pro now in main model picker
- GPT-4o is now also available to Plus and Team users. To use it across platforms, go to settings on ChatGPT web and toggle on “show legacy models.”
10
979
It is never a good idea to not offer a neutral fourth option, as people who do not have an opinion but want to see the result will press anything 😬
1
13
933
Wow, cool! It would be great if tasks worked with projects. Then I could say: Create a new Spanish exercise every morning at 09:00 and base it on our learning progress.
3
1
12
1,581
Hmm, I think it’s mostly positive energy tho. So much bad shit on this platform, I kinda like the vagueposting; feels a bit like before Christmas
13
3,728
Wow, that's awesome! Please also give us the ability to attach documents that are relevant to the writing (e.g. writing style, brand positioning, product information, USPs ...)! It's annoying when you have to upload the same 10 documents for every new chat.
1
263
agree:
honestly gemini's glacing is even worse than chatgpts, it's crazy. a few examples from the first lines of four CONSECUTIVE chat messages today:
This is a phenomenal find and you've absolutely hit the nail on the head.
Yes, that's an excellent idea! It's much stronger visually and in terms of content.
Yes, absolutely! That's a crucial point and an excellent addition.
Yes, that is a very good and important point.
---
pls i love gemini and 2.5 pro a lot, but the glazing and the way gemini only gives positive feedback is really bad for work stuff.
cc @OfficialLoganK @joshwoodward
11
1,259
Why don’t the follow-up with clear communication then? Would have been easy to fix. The backtracking suggests their reasoning isn’t strong. What do you think, @GaryMarcus?
2
9
7,129
there are so many errors in this post; it's crazy
"And last week it released its browser-using agent Operator to the public." operator happened in january i think? this is a new agent
"It now claims that more than 400 million people use its products every week" it's 700 million WAU now
“[GPT-5] is an experimental model that incorporates new research techniques we will use in future models,” he was referring to the math model
there are even more lol
1
12
5,889
"All files uploaded to a GPT or a ChatGPT conversation have a hard limit of 512MB per file. For images, there's a limit of 20MB per image."
help.openai.com/en/articles/…
12
314
you guys are obviously leading with creative vision in the video space; my only wish is to get better video quality, still getting better raw quality with veo 2 than gen-4
1
12
1,168
Here we go. The @BoredApeYC has just hinted at a collaboration with one of the world's biggest sports brands - @adidasoriginals.
Following the announcement that Adidas has purchased land in @TheSandboxGame's #Metaverse, this is a logical next step for the brand.
1
3
11
Because there won’t be an open web anymore. Everyone‘s putting up paywalls - see Reddit selling API access to Google and OpenAI. Or OpenAI making deals with all those media companies to have access to realtime data. xAI has a huge advantage to have free access to realtime data especially for queries that are past the training data cutoff. Don’t you think?
4
1
10
409
agree. i feel like many many people on here need to stop adjusting their beliefs after every launch; it's actually crazy how much of a rollercoaster most people's beliefs are.
also too many folks are far too emotionally involved
10
915
that’s a great take; I’m exploring the model since a few hours and it’s definitely something else and the value prop for the price is questionable but it’s for example by far the funniest model I have ever used. 4.5 is definitely one of the weirdest releases in a while. gotta respect that.
1
9
1,119
In den Niederladen gibt's das auch. 5 Jahre lang bekommt man Steuervergünstigungen und anschließend kann man einen niederländischen Pass bekommen. Ich kann mir gut vorstellen, dass sich diese Systeme langfristig rechnen.
2
7
774
you think it's a good idea to listen to feedback here on x? seems like it's the arc audience (myself included) who ask for power user features (again) like vertical tabs, spaces ... how do you handle this situation?
1
10
3,354
The potential of Gemini in Workspacs is nuts. I would love a better email integration. Right now, the writing quality, context understanding is much better when I copy and paste the email conversation into the Gemini app and write an answer there compared to using Gemini in Gmail. Would love for this to change because it doesn’t make sense to copy and paste to the app.
1
9
1,038
I spend time in the Help Center almost every day. The Help Center is great. But it also shows the incredible complexity that ChatGPT has built up in the meantime.
help.openai.com/en/collectio…
1
9
1,820
do you think it matters? i think people are massively overvaluing model dicussions. it's a commodity by now. the value is mostly in the product and chatgpt just crossed 400 million WAU
1
10
865
It’s the weirdest model release since a while. Cost/benchmark performance is ridiculous but at the same time it’s probably the most addictive and also funniest model I ever tried.
8
1,782
i am going to laugh my ass off if this is just going to be a disney collab announcements (there were rumors i think)
1
8
648
agree; the main reason many people on here talk shit about gpt-5 is because they expected a revolution compared to o3 which i think is a crazy thing to expect considering how much of a jump o3 was
9
517
I understand that your vision is bigger than just serving power users, but I would like more clarity on what is happening to us loyal Arc users and whether we should be looking for something else.
7
1,554
Sitting on the OpenAI board and praising their biggest competitor in a paper, then firing the CEO and approaching said competitor for a merger seems a bit weird tho?
2
6
1,032
We’re at a point where the beta testing name is better than real model naming 😅 (I think it’s great you did this!)
8
1,032
Does this do the same as picking 5 Thinking? Or should I say ‘think hard’ with 5 Thinking too?
2
7
3,403
