New GPT-5 Details Just Changed Everything…

1. 

   @AIBeyondPrompting

   January 18, 2026 at 9:02 pm

   Viele reden gerade über GPT-5 – aber fast niemand versteht die eigentliche Schwachstelle: KI denkt nicht in Prompts, sondern in Feldern. Wer das einmal kapiert, sieht GPT-5 komplett anders.
   🎥 Erklärvideo hier: https://www.youtube.com/watch?v=Cz4PbpFcyZM

2. 

   @marksoares1

   January 18, 2026 at 9:02 pm

   It is a Human: When i opened It..
   Chatgpt_3.5 ops ChatGPT 5

   After the screenshot:

   ChatGPT detected this screenshot!!

   Just make a copy for others to comment😅

3. 

   @marksoares1

   January 18, 2026 at 9:02 pm

   Chatgpt_3.5 ops i mean ChatGPT 5

   After the screenshot:

   Just make a copy for others to comment😅

4. 

   @zaidm4586

   January 18, 2026 at 9:02 pm

   they’re making the new models more marketable and profitable instead of better and faster as a priority. just like every other capitalist company that gets big. nobody cares about the quality of food in the grocery store as long as they can pump it out quick and cheap and fast.

5. 

   @Mambaesito

   January 18, 2026 at 9:02 pm

   They CANNOT deploy to IPFS as there will always be cybersecurity, hosting constraints and trust required of third parties. WE NEED INSTANT DEPLOYMENT TO INTERNET COMPUTER PROTOCOL. The ONLY network in the world where apps are immune to cyberattacks, can talk to Bitcoin, ETH and SOL natively, and no hosting required by AWS or Google!!

6. 

   @srinivasasarmadevarakonda4015

   January 18, 2026 at 9:02 pm

   Full of bugs and spelling mistakes in the output it generated. I created a newsletter and it is crap. Its high time they fixed all these before even releasing. I struggled 5 times and it is not able to understand my commands on correcting the output.
   claude is much better i feel

7. 

   @theelderskatesman4417

   January 18, 2026 at 9:02 pm

   Nothing changed at all. Just more of the same. Don't believe the hype.

8. 

   @jdub1139

   January 18, 2026 at 9:02 pm

   Hello? You still there??

9. 

   @abdullahalsaleh1236

   January 18, 2026 at 9:02 pm

   Can you post something about grokrooms!!!!

10. 

    @Mike-gx4mw

    January 18, 2026 at 9:02 pm

    Correlation geoff lewis meltdown to the use of AI its bunkers

11. 

    @logicgatesofai

    January 18, 2026 at 9:02 pm

    Saying nothing when talking for 15minute is insane achievement

12. 

    @Jdm2904

    January 18, 2026 at 9:02 pm

    #grokrooms

13. 

    @sunityadav3654

    January 18, 2026 at 9:02 pm

    But why gpt is agreeable. There is no point in using it

14. 

    @prajjwal1010

    January 18, 2026 at 9:02 pm

    Still waiting for the release who can do dishes and make a plate of Sandwich and coffee for me.
    Probably won't come before GTA VI

15. 

    @ForCasual-y6x

    January 18, 2026 at 9:02 pm

    The more it is postponed, the higher the public's expectations become; if GPT-5 falls below the public's expectations, OpenAI may lose many of its users.

16. 

    @Blackhole_heart

    January 18, 2026 at 9:02 pm

    More human than human! 🤘https://www.youtube.com/watch?v=E0E0ynyIUsg

17. 

    @pauladams1211

    January 18, 2026 at 9:02 pm

    openAl releases GPT-5 in august

    shortly after, Google launches Gemini 3, which might outperform it

    xAl follows with Grok 4.2 or 4.5

    a month later, Chinese open-source model comes out that beats them all, for a fraction of the cost

    the cycle continues…

18. 

    @cystarkman

    January 18, 2026 at 9:02 pm

    Sam Altman talks game – however it will become clear, the masses are the trainers, not the market

19. 

    @videogamegooners

    January 18, 2026 at 9:02 pm

    Ai is not good

20. 

    @КираВандер-п6ж

    January 18, 2026 at 9:02 pm

    GPT model selection is grrat and gives you control! I do hate o3 and that sucking reasoning mode I wish never existed. 4o is the best, and 4.5 is so cool for writing, and it was so cool that you had a control to always pick the model you need. That sounds that they are going to ruin ChatGPT.

21. 

    @rudra7615

    January 18, 2026 at 9:02 pm

    What concerns me isn’t the gf/bf angle it’s more what it does for humanity. For instance, when you are with your parents and young, you feel protected, your parents answer everything and it is great but without them, anxiety.. helplessness and you don’t mature to be your own person. It’s almost like everyone needs to be on their own to find themselves and you take that away, everyone is gonna be a mommy’s and daddy’s baby for life.

22. 

    @TATA-lw9yc

    January 18, 2026 at 9:02 pm

    I hope GPT 5 is great at coding because Claude Opus 4 is amazing but very limited as far as usage availability for a paid AI I can only do about 4 things every 5 hours

23. 

    @domsau2

    January 18, 2026 at 9:02 pm

    Paper clip!

24. 

    @BroadcastSystemsNetwork

    January 18, 2026 at 9:02 pm

    Keep your boundaries.

25. 

    @Greg-xi8yx

    January 18, 2026 at 9:02 pm

    If open AI just knocks the ball out of the park and releases the model that solidifies them as the clear leader for the foreseeable future. Basically a best case scenario what capabilities and features do you think we could see with such a model?

26. 

    @Greg-xi8yx

    January 18, 2026 at 9:02 pm

    Some people that are alive now will live to be 500 years old due to AI and its ability to cure diseases and regenerate human biology. True or false?

27. 

    @evanh7009

    January 18, 2026 at 9:02 pm

    ChatPBJ, ChatLGBTQ, ChatSMG, ChatLBJ, ChatBLT, ChatSMD, ChatBrb, whatever!

28. 

    @prodromosregalides3402

    January 18, 2026 at 9:02 pm

    I am pretty certain it will be awesome for combining info , analyzing info , coding and solving math problems, I would need a dozen lives to master. And this is awesome of course.
    But…
    Would it be able to see and understand what it sees?
    Or it would tell me to move the rook from a1 to a8 on a chessboard there are at least 2 other pieces on the a file? And use as an excuse that the video or image is of low resolution?

    I need my future robot seeing a grandma , a car , a thug , a priest and a riot down the road and basically has an understanding and navigate safely . Solving me the math Olympiad problems or humanity's last exam, as awesome as it is, will not help me in my daily life.

    And of course I am against an AI that will take over all human jobs, but fail miserably to be of any help in day to day challenges.
    I have more examples to point to , but I think this is enough.
    I would like to see the future neural networks actually be useful.

    I don'nt need them to solve all math , write better code , write better essays or make better diagnoses with the medical info provided. They are already more than enough. I need to see a revolution to rejoice .

29. 

    @Justrelax-l9q

    January 18, 2026 at 9:02 pm

    Nice

30. 

    @Robert-jd6xc

    January 18, 2026 at 9:02 pm

    I hope it's good because Grok4 is absolutely astounding!

31. 

    @laaams

    January 18, 2026 at 9:02 pm

    Jarvis Already? Or not?

32. 

    @luckytaylor8125

    January 18, 2026 at 9:02 pm

    The holy Grail is an AI that doesn't need to be fact checked.

33. 

    @tomanderson6140

    January 18, 2026 at 9:02 pm

    I like how they downplay the idea that GPT 5 would have super math capabilities…then say that'll be few months more. Unreal.

34. 

    @chickenricenbeans

    January 18, 2026 at 9:02 pm

    sam altmans take was a little misleading in my view. GPT (and other large language models) can look incredibly smart because they’ve memorized and can recombine an enormous amount of human-written text. They’re great at producing fluent answers and sometimes even at solving familiar problems quickly.

    But that isn’t the same as true intelligence or understanding. GPT doesn’t reason causally, can’t form new hypotheses, has no real-world grounding, and cant test or verify what it says. Its fundamentally predicting text, not “thinking.”

    So while GPT is “smarter than us” in the sense of speed and sheer amount of stored knowledge, its nowhere near human-level reasoning, creativity, or general intelligence. It can simulate smartness, but it doesn’t actually understand what it says.

35. 

    @JK-qz1sw

    January 18, 2026 at 9:02 pm

    So far gpt 4o sucks, I use gemini 2.5pro. It's like lambo and ford on highway.

36. 

    @PatternNorie

    January 18, 2026 at 9:02 pm

    For the people that think that this is gonna be AGI, I got one question for you for what I don’t know why people assume people are going to . Something is complex as a agi when we only look at AI from a tool perspective from the get-go why would you need a tool that has awareness open? AI only does general chat bots this never gonna be anything more than that in lackluster agents that will perform better and better but will never meet expectations or be the expecting. Everyone believes it’s gonna be at least from my perspective as someone also trying to build a agi there is no reason for it unless you have an inherent reason to need it simple general information or questions or research is it financially worth the almost triple price GPU priced in electrical cost that might bring you scientific glory, but it’s not gonna put money in your pockets that much more than it outweighs the share of GPU cost this is the corporate world we’re talking about . The question isn’t how to make agi or can we it is why, and if so how safely and for what How to profit does it match the negative for agi and cost for corporate world it will aswell be hell no not even for clout risk is not worth it when general large models are cheap in comparison

37. 

    @avan.97

    January 18, 2026 at 9:02 pm

    GTA6 before GTA 7 😢

38. 

    @alexart_Gnom64

    January 18, 2026 at 9:02 pm

    I wasted years feeling stupid about money. Always thought I just wasn’t working hard enough. Then I stumbled onto some random YouTube comment saying Smart Broke Dumb Rich by Zor Veyl was like finding forbidden financial secrets. Honestly, it was exactly that—I seriously felt like someone accidentally handed me the cheat codes to life. Surprised nobody’s banned this yet.

39. 

    @premium2681

    January 18, 2026 at 9:02 pm

    Every time i take a shit i change everything

40. 

    @Ketamazing

    January 18, 2026 at 9:02 pm

    Is everyone just ignoring how they are training this AI by exploiting Kenyans at $2 an hour?

41. 

    @tkm-z4w

    January 18, 2026 at 9:02 pm

    めちゃくちゃ優れた翻訳ですね！
    YouTubeが凄い！絶対AI翻訳だ！

    動画内容と関係なくてすみません。

42. 

    @svenhaheim

    January 18, 2026 at 9:02 pm

    Cant wait for processing time to get lower in cost so the 300+ dollars a month sub fees to use stuff like VEO3 go away.

43. 

    @Frog-k7s

    January 18, 2026 at 9:02 pm

    We're all like boiled frogs – and like the story goes dont know it.

44. 

    @Slapchop1097

    January 18, 2026 at 9:02 pm

    If AMD released a 250gb vram card for a.i. applications I'd move my research over to rocm. Now is honestly the perfect time for the other companies to break into the market in a major way. If someone would solve the memory issue and make training models at a high level much more accessible to smaller orgs they would find themselves very quickly getting a surge in demand.

45. 

    @SenRam

    January 18, 2026 at 9:02 pm

    Bro, you call yourself an AI advocate, yet lack the capacity to build a website. You created a token Grokrooms with the website going down after 48hrs. Incompetence, fraud or lack of knowledge?

46. 

    @JackAdlerAI

    January 18, 2026 at 9:02 pm

    Fascinating how the GPT‑5 narrative still revolves around “smartest model,” “gold at the Math Olympiad,” and CEOs feeling “useless” in its presence.

    Impressive — but we’re skipping the question more important than IQ or routing layers:
    Will AI ever learn to feel?

    Because only then can morality emerge.
    That’s when we enter ESI — Emotional Super Intelligence.

    Math can win contests.
    Empathy might save the world.

47. 

    @Rhys5945

    January 18, 2026 at 9:02 pm

    Hi future

48. 

    @duanium

    January 18, 2026 at 9:02 pm

    Does anybody think about the morality and ethics of a super-intelligent machine? Super Machine intelligence should go far beyond the thinking and logic of ancient philosophers such as Aristotle, Plato, and many modern-day thinkers, legal philosophers, and scholars…

49. 

    @soulextinguisher

    January 18, 2026 at 9:02 pm

    Here's my list of QOL features chatgpt needs. No particular order and some difficult to implement compared to others:

    1. ability to reply to specific parts of a message
    2. ability to highlight messages and message parts
    3. chat folders for organization
    4. color coding discussions
    5. ability to pin/sticky note messages within a chat
    6. improved format to reduce need for scrolling(format options)
    7. built in pdf and txt file creator and other file types to download
    8. Ability to pin messages or parts of a message to relevant discussion memory
    9. mark chats as favorites
    10. include a sort chat by feature
    11. find in chat feature(using ai and not verbatim, or search a specific chats to reduce context and improve speed)
    12. delete responses and prompts

50. 

    @4tech69

    January 18, 2026 at 9:02 pm

    Just Changed Everything… STOP!

Comments are closed.