Universal Memory Tool

5 days. One prompt: "Summarize each stated fact — who did what, when, where." Still can't do it. Wrong dates, skipped facts, half the speakers confused. This is the model everyone says is smart enough to replace analysts. #ChatGpt $100 sub, reasoning effort on high. If you need proof, i have plenty of screenshots 😂

@YashHustle_22 First it kept ignoring my prompts, now it stops after every session. Slowly i feel scammed. #ChatGPT

What's your main AI model right now? - Codex - GPT 5.5 - Claude Opus 4.8 - Claude Sonnet 4.6 - Gemini 3.1 - Claude Opus 4.7 - Grok 4.3

@YashHustle_22 They all suck eventually 😂

@sama It can't even finish a simple task, how is it supposed to code a whole website?? 😂😂😂 5 days wasted on a $100 sub.

build and publish web apps with chatgpt! i really wish i had this when i was a kid, but i do miss hypercard.

This can't be real. #ChatGPT How often do i have to repeat myself ? 😭😭😭 These devs really shouldn't get paid that much money for some useless model.😂 Like it can't be that dumb?

Why? seriously why? Fix reasoning in #ChatGPT 5.5 please! I can't work like this! $100 sub, reasoning effort is high. And it can't finish one simple request for 5 days now!

@OpenAI why is ChatGPT so dumb? Please tell me? 5 days and not one usefull summary!

@OpenAI i want my money back! 😭😭😭😭😭 4 days of #ChatGPT trying to avoid it's task.

Can you please fix #ChatGPT, no matter how often i tell it to read and summarize it does something else. #total #fail #ai #gpt 4 days that i am working on this problem. Long-term memory engine works perfectly fine. Your models are not capable enough, that's it..

Okay so the creator's of the #LoCoMo test for llms should really reconsider their grading logic. Answer: their dad get's less points. Gold F1 answer is Dad Next question: doesn't contain "and" but has both names, less points. Come one? Did you read and test your scoring ? 😂 In the past 3 days i have found so many mistakes and flaws.. #ai #test #memory #reasoning

Like seriously? Why would @chatgpt write a candidate limit of 2000 characters into my search engine without being told to? With a database bigger then 5GB and reasoning tests with more than 10M tokens. Why would you do that ? If the original code never had a search limit?

Hey @sama fix your models please!

@X stop ignoring my posts😂 like I said. I don't know why llms are so hyped. i can give idk how many examples where #claude and #chatgpt are failing horribly.

I don't know why everyone is praising GPT-5.5. I pay $100 a month for a model that ignores instructions and forgets half of what it's supposed to do. @sama Running the #LoCoMo test with your models is horrible!! The highest score I got on F1 was 0.53 with my memory engine. LLMs are too dumb to figure out the difference between third and first person. Question: What was Melanie doing in 2023? Test text clearly says speaker=Melanie, text="I went to the beach ." 😭😭😭 LLM query: Melanie, doing, 2023. Even with "soft rules" and no hints, your model is not able to figure it out. #GPT #ai #dumb

@Deepvalueagent Building TesterBuddy, connecting devs with testers for direct chat feedback. iOS: apps.apple.com/app/apple-stor…, Android in public beta — DM me!

@dawedeveloper Sounds very useful. Sadly we will be on dekstop only for a while...

What are you building today? \(•_•) ) )z / \ Comment below 👇

@mytwillot That sounds really interesting, now that you mention it, I remember that i have some bookmarked posts i wanted to read but forgot about. 😃

@Exposes_ Hey Nick, let's connect

🚀 Looking to connect with: • AI Builders • Software Engineers • Startup Founders • Indie Hackers • Students learning tech I'm sharing AI tools, startup lessons, tech insights, and exposing the hype. If you're building something cool, drop a 👋 below and let's connect. The next big opportunity will come from the people you meet today. 🤝 #AI #Startups #Tech #BuildInPublic