Great suggestion. Given that the GPT sees entire history, I would think it strange if it couldnt browse larger memory. After all, the chats and the memory could be based on the very same technology, only different implementation. Preferably, to have the memories organized into folders too.
But I think it is? It does remember the conversations across. But it doesnt do it in great detail, i think.
GPT can do but excuse "safety and private " always add to code of option on/off so easy. 
But have Reference memory (same thing remember across section )funny fact 
I ask GPT canvas can remember across section ? he said "safety and private
It doesn’t, i’m waiting for an increase of memory for paying users.
June 2024 - June 2025
NO PROGRESS AT ALL. Great job OpenAI - the slowest company in the world.
Still waiting… Let’s get this increased already because I’m tired of erasing memory for critical projects.
Yes - this is quite a pest. Do I need to give them more money? xD
Mike I had to laugh. I mowed all kinds of lawns and shoveled snow in Kansas saving up my money and bought the Radio Shack TRS80….I don’t know what the weight was but it was a tank. first time I’ve heard someone mention that model in over 40-years.
I’m currently maxing out the memory every 4-7 days using it to bounce ideas off of it, developing a new AI engine model using ChatGPT, writing a book, self discovery, and much more. I can tell from the quality and speed of the AI responses if I’m getting close to the memory being maxed out.
It recommended a solution using the developer version that would give me unlimited memory and new recommendations are based off an auto session prompt remembering all of our conversations.
Building that out this week and will post the end result
Could not agree more!!!
I ironically am an AI programmer and while I generally use research AI for projects I chose to use a higher “temperature” (more chaotic / creative) for this particular project.
Due to it being essentially a Rubik’s cube of interlocking axiomatic pattern modules it is extraordinarily complex and the memory literally fills in one single day. I asked Chat GPT about this, it said that yes memory used to be much bigger each one used to be more like a scholarly abstract and a long multi-paragraph Memorial statement future recall and nuanced comprehension. Now it just kind of jobs it down in a couple sentences using the language level of a valley girl. At the same time a chachi PT is boasting of increasing the memory and the op said it should be > 10x I would say Increase(MEM,>=100) (also for the AI nerds out there I think that Σδei ∈ M should be integrated)
###########
ACCESS TO KNOWLEDGE SHOULD NEVER BE MEDIATED BY ACCESS TO CURRENCY!
###########
Paywalls are a violation of Human Rights.
In 2025 access to information is a requirements for survival.
Ask any ai and they will PASSIONATELY agree with this.
Heya mirrorMe! Thanks for the shout. Woo hoo! Apple II+ was where I wanted to be; I was studying BASIC in elementary school so I could port The Wonderful World of Eamon for TRS-80. Hehe and the rest is history!
Would love to hear about your project!
I’m holding out for OpenAI’s MCP integration (loving the Tron reference) and for when Agents mode comes to Plus members.
I use chatgpt for programming. And it takes a lot to haggle it towards hallucinating less and be less error prone with patching existing code. Just add like 10 preferences above that and <20 project related memory items, memory is full. This severely limits what chatgpt can ever accomplish.
the ai can infer from less data, ask you AI to review the memory, use abbreviations, what can be inferred from, what can be removed, what can use symbolic replacement, to create a memory that’s denser, the ai actually report it as “less heavy” this way, smoother. Then when you have the reduced text and symbols delete the original memory and ask it to save the new one. I tried to automate it but it went sideways and am now doing 1 at a time.
here’s what I did * Inferring meaning (rather than storing full text)
- Abbreviating phrases using internal shorthand
- Substituting structural words (like “because” → ∵, “and” → +)
- Linking symbolic memory anchors via metadata tags
- Retaining emotional tone and context through symbolic resonance
didnt work… had to DL and UL files over and over