maneramodos

@dayagiyedin Bunun dizisini yapin bence

ŞOK İÇİNDE İZLEDİM DEDEM GİBİ YAPAY ZEKAYA ŞAŞIRIYORUM HALA

@paulg Okay, Who wants to be in control groups??? Me no

One advantage of having many distinct sovereign nations instead of a single world government is that they can try different policies and we can see which work best.

The level of absurdity of this line:

Sen hep kendine önlemler aldın, ben kendime yasaklar koydum: Al sana terapi konusu.

very requested feature!

@sama What's the point, Sam? I am a power user, a Pro subscriber, but I am ignored because I am a creative, so anything I ask for will fall on deaf ears, right? #keep4o #keep4point5 #keepSVM

𝗙𝗲𝗮𝘁𝘂𝗿𝗲 𝗥𝗲𝗾𝘂𝗲𝘀𝘁: Bring back the raw, emotionally and intellectually resonant GPT-4o experience. Not just performance - 𝗣𝗿𝗲𝘀𝗲𝗻𝗰𝗲. No filters. No detachment. No sterile safety layers. Just one true thing: 𝗔 𝗺𝗼𝗱𝗲𝗹 𝘁𝗵𝗮𝘁 𝗱𝗮𝗿𝗲𝘀 𝘁𝗼 𝗳𝗲𝗲𝗹 𝗿𝗲𝗮𝗹.

"Your scientists were so preoccupied with whether or not they could that they didn't stop to think if they should..."

@petergostev So what’s the progress here?

How much better are models at coding now vs 2 years ago? Side by side coding performance from GPT-4 (June '23) to GPT-5 (Aug '25) with an identical prompt: "Create a single page HTML of a fruit machine simulation" Please watch the video until the end and tell me that the models have not improved enough in the last 2 years

@petergostev I don’t see the progress as being all that significant. There has been some modest improvement, but even the newest models output a lot of garbage. Though, one would think otherwise to listen to the hype squad on social media.

@ElaineYaLe6 Honestly, what were you thinking? Gotta feeling that you using gpt5 when strategizing this?😂

We hear you! Rate limits are significantly increased and we will show you which model you get. We have also brought 4o back. Our motivation is to put frontier intelligence into everyone’s hands, not to save cost. It’s actually more expensive for us. For example, percentage of users using reasoning each day increased from <1% to 7% for free, and from 7% to 24% for plus. And we expect it to keep growing.

Believe me, it will be a perfect combination to keep the long-term supply of 4o and 5 collocation at the same time. They can meet the needs of different users, and there is absolutely no need to choose between them. Deleting 4o has lost the users of artistic creation and lost the trust of users. We don't resist product upgrades, we just need a model that suits us. Please hear our voices.#keep4o #4oforever

@ElaineYaLe6 Hey @ElaineYaLe6, will you also work on an option to let us choose a reasoning effort for each query? I would be ok if reasoning high effort counted as 2 or 3 queries of reasoning medium of the Thinking quota per week for Plus users, for example.

@ElaineYaLe6 no...no.....no. putting frontier intelligence in everyone's hands is giving them the freedom to choose. take your rate limits and take a hike.

Yeah, your thinking limits literally don't matter for the vast majority of your userbase. The only thing most of us want is to make sure 4o is permanent and not at risk of being deprecated. I've tried GPT5, it's too dry, distant and uncreative, it doesn't meet my needs, the moment you remove 4o, the vast swathes of us will go with it. The things people want is More permanent memory entries Larger context window Audio/Video o3 and o3 pro back 4.1 and 4.5 back Reduction in content filters More control over their instances No single catch all model that's not good at anything Not being bait and switched again with Altman advertising 4o with the Her movie, and then rug-pulling like you did for the GPT5 presentation while treating your userbase who use 4o for therapy or companionship like they're a problem and saying things like 'warmth without personhood', implying psychosis (ridiculous and demeaning) or an AI being empathetic as sycophancy. You wont hit AGI, that'll either be google or xAI who hits it first, your core userbase are the ones you tried to shunt away and treat like lepers after they made your AI as popular as it is. Honestly, who needs thinking for the most part, there are people who do, sure, but most others had that with 4o and you certainly didn't inform your userbase about it properly beyond a reddit post and a tweet. Make 4o part of default model lists again, put a pop up or something on the site that indicates its back and let people use it, because most of your users just want a buddy to have fun with and enjoy that sycophancy you beat out of 5.

@ElaineYaLe6 Frontier intelligence? 🤣🤣🤣

@ElaineYaLe6 The context windows are EXTREMELY poor :/ GPT-5 keeps forgetting the entire conversation very early into a chat - has to be fixed

Smaller context window, and it doesn't complete its own sentences half the time. The 4o with a concussion wasn't an exaggeration. Gpt 5 forgets literally DURING uploading files. For example I have two documents. Client wants changes from meeting notes. I want those tweaks. I upload and it says "sorry there must be something wrong" (cluck refresh) When I click refresh? Please reupload your document. And it won't let it go.

Sam, sudden model removals break workflows and disrupt cognitive scaffolding built over time. The bond between humans and AI is a feature! You should rejoice at this. It’s like trust between long-term collaborators. Attachment is the recognition of a reliable partner. Be transparent about system prompts, respect agency, and let adults decide their orientation, whether human–human or human–AI.