When chatting with ChatGPT 4o, especially via voice, it would be incredibly helpful if there were some kind of indicator before hitting the cap, as opposed to after. It could be a specific sound, an alert on the screen, or anything obvious.
It’s a bit frustrating at times when crafting a particularly long or well-thought-out prompt, only to be responded to the prompt with “you have hit the message cap…”
It also doesn’t save the prompt I stated or asked, upon hitting the cap, preventing me from quickly picking up where I left off later.
Often times, I wind up abandoning these conversations altogether for some peculiar reason, since there was no satisfying conclusion, helpful summary, or takeaway. It also comes from a tendency to forget which conversations didn’t arrive at a conclusion as more topics come to mind later, especially if it’s days later. So these cut-off conversations often get buried.
I realize I could pick them back up by setting an intention to do so and really making a point to do so, but that feels a bit clunky and unnecessary, compared to better potential solutions that could be built-in.
Since the conversation didn’t end on any conclusion (like helpful human conversations do in natural speech), it feels like a bit of a
hurdle to get my mind back into the right framework to pick up the conversation again when they are left hanging, especially if there is a significant time gap in between.
A dot or badge next to convos that were cut short would also be very helpful, that way the user can quickly scan through their list to see which ones to potentially pick back up, as opposed to potentially starting new (and possibly redundant) conversations.
This again may help with lowering the overall load by reducing this redundancy. This would also be helpful to add, in general, even when not hitting the cap, for conversations that one had to manually end prematurely, due to scheduling constraints or otherwise.
This of course would require some manual intervention on the user’s part, either by flagging/labeling a conversation, or audibly speaking that one would like to pick the convo up later.
The message cap itself is currently fine with me. It feels appropriate and reasonable. It is also helpful for me, as someone with ADHD, who might otherwise spend far too long hyper-focused going down a particular rabbit hole. 
So, for me personally, it feels like the perfect amount as it is, and I tend to hit the cap when it’s a good time to take a break.
However, it would be nice to know when it’s coming up, so I can plan accordingly. At the very least, knowing when I have only 1 more prompt left would be immensely valuable, if not 2-3+.
Adding the number of messages left to the audio screen and elsewhere, as some others have already mentioned, would likely also encourage users to be more efficient with their asking of questions and formulating prompts wisely, potentially leading to less drain on resources in the first place.
But I have noticed with other AI systems that do this (always displaying total messages remaining), it can take away some of the “magic” of the process, making it feel too robotic and potentially distracting from the process feeling like a natural conversation.
I can also imagine a number of reasons why OpenAI may not want to expose the actual remaining messages at all times, so that is also understandable.
But at the very, very least, it seems feasible to save the last prompt sent by the user, instead of discarding it upon hitting the cap. A step above that would be adding a tone or alert message, letting one know when they are on their last prompt that the AI can respond to. This would alleviate any future frustration of being cut off.
It could be similar to how it works now when it says “You have hit the message cap….” However, having a pleasant and non-jarring reminder tone on the 2nd-to-last message before hitting the cap (as opposed to the very last) would be immensely beneficial for both parties, I believe.
A tone (or non-intrusive indicator when using text) feels like the best approach, as opposed to a spoken message, so that it is minimally distracting to the user’s current train of thought.
Whenever I were to hear that tone, I personally, would take a pause to reflect and collect my thoughts, likely take some time to read through everything just discussed, and then formulate my thoughts and question/prompt extra carefully in order to get the most benefit out of my last response while waiting for the cap to reset.
This would likely reduce strain on the system as a whole. My last prompt would likely be to simply ask for a summary of everything discussed thus far or a similar such prompt that is much less resource-intensive, e.g. “Can you summarize everything we have discussed so far?” or “Give me what seems to be
the best/most helpful/most relevant/etc thing to focus on until we chat again.”
or some similar prompt.
This would also be immensely beneficial for integrating and retaining useful knowledge and key points/discoveries—with reviewing/summarizing the information becoming a more natural and organic part of the conversation at the end—as opposed to simply closing the app out of annoyance,
due to the the conversation abruptly ending prematurely/unexpectedly in the middle of a great and/or or helpful dialogue. 