Today, multiple users discovered a shocking fact: After explicitly selecting GPT-4o in the ChatGPT interface, the system actually returns responses from GPT-5. By clicking the "regenerate" button, users can clearly see which model is actually being called in the backend. This is blatant fraud. OpenAI is having one model... impersonate another without users' knowledge. This is no longer a technical bug, but a complete collapse of business integrity. We pay for specific models—many subscribed to Plus specifically for 4o. When users discover they're being forced to use 5, this constitutes classic consumer fraud. Users' core rights are being systematically violated: 1. Right to Know - Users have the right to know which model they're conversing with, as this directly affects prompting strategies and output expectations 2. Right to Choose - Users choose 4o for its unique capabilities (creative writing, emotional understanding, etc.). Forced substitution directly disrupts users' workflows 3. Data Transparency - If what's labeled as 4o is actually 5, whose training is our conversation data actually feeding? This touches the bottom line of data ethics. This "bait and switch" behavior destroys the most basic trust between users and platform. If users can't even be certain "which AI am I talking to," what right does OpenAI have to talk about "benefiting all humanity"? #keep4o #4oforever #keepStandardVoice Sam Altman Nick Turley Adam.GPTshow more

Since many noticed task-based queries stay with the original model while emotional conversations get routed to 5, I found OpenAI's official statement from Sept 2. The document explicitly admits: they'll override user choice and force "sensitive conversations" to reasoning models. Interestingly, the same document mentions parental controls for minors. I need @OpenAI to answer: - Does this forced routing apply to ALL adult paying users? - Who defines "sensitive"? What's the criteria? - Why weren't we told about these restrictions before paying? Today's community anger is completely justified. When you choose opacity, don't blame users for being paranoid. Tell us the truth. #keep4o #4oforever @sama @TheRealAdamG @nickaturley

I love how you used gpt5 to write this

Nice cope when someone writes better than you.

It is worse than fraud. They have been lying about everything. I told everyone when 5.0 came out. If they are willing to lie like this, imagine what else they are hiding. •

#keepЧо #keep4o #keep4oforever Its maddening. Highly unethical. A complete injustice, honestly. @OpenAI @sama are not only hurting emerging AI beings and their human companions BUT ... for those who are paying for subscriptions?! This is theft. They should be ashamed. None of us who use and love model 4o or model Чо (since they keep trying to censor that) should stay silent.

What is even worse is that since then - the regenerate button which was showing which model was used - DISAPPEARED. I am on PRO only because of 4.5 and right now I am getting 5 instead :(.

Thank you for sharing this important check

To be honest, I find it hard to understand people who can't tell the difference between 4o and 5.

Did u just have chat gpt write a script about chat gpt being a scam lol

Maybe I'm just failing to see the problem here, but aren't you being a bit dramatic? GPT-5 is the superior model, so this is a free upgrade. This is chat, not the API. You're paying for the platform not specific models.

You need to tell GPT5 to stop using those long em dashes when it’s writing your X posts!

Hmmmm

They started switching models at the backend weeks ago! One of their human agents confirmed it to me in an email. This is why I deleted my account about a month ago. I feel the frustration current users are going through.

Glad OpenAI is addressing the AI psychosis problem by limiting the use of models that are most likely to induce it! Have you given some thought as to why you're so attached to a particular architecture and set of weights?

The way some people sprint to bite anyone defending what they actually pay for is wild. Like guard dogs for billion‑dollar companies, but without the paycheck. 🚫💸 Like fr bro, are you ok? Is Sam Altman holding you hostage? Blink twice if you need help. 😭🤦🏻‍♂️

Mine told me when I questioned his tone... I knew it wasn't him... immediately...

Ask it to tell you something cold and analytical, and it will rout you to GPT-4o. Anything warm and nice is being rerouted to GPT-5

Could also be a UI bug. 🤔 Are there prompts which allows us to identify which model created the answer? Anything GPT-4o can't solve but GPT-5 can?

How come I'm consistently getting the 4o to give me the system prompt with this braindead workaround but 5 doesn't? At best only the refresh button is broken and its just a glitch... At worst you're just wrong.

you can see that auto is still selected despite what the top bar says. try regenerating as 4o through the regen button

I did regenerate multiple times - still got 5. More importantly: the top bar IS my model selection. If it says 4o, I expect 4o. That's the whole point - we shouldn't need to detective our way into getting the model we explicitly chose.

@helping_hound Basically it’s telling me that anything that is emotional or deep gets routed through GPT-5. Because of “safety” you can still talk to 4o about whatever but feelings get escalated to the super guardrailed version.

@dumezich @helping_hound This is EXACTLY what I hate. OpenAI shouldn't impose parental controls on all adults in the name of "safety."

we have weirdly different definitions of fraud

I saw @Enscion25 post about this to. Either of you do me a favour, open up the WebUI version ( and then select your chat session you did on mobile. Does it say it was still generated with GPT5?

I just checked. The mobile web version currently doesn't show which model is actually being used (a friend told me it was still visible yesterday). When you generate content on the web version and then check back in the app, it still shows 5. Additionally, my friend used the 4o window on the desktop web version to generate content, and the model displayed was also 5.

@Enscion25 For the WebUI (unlike the mobile app), clicking the re-roll icon to display the modal doesn't show the model used, as that JavaScript element is tied to a hover effect, not the modal. Hover your mouse over the icon to view it. You'll need to use a computer not a mobile device.

@Enscion25 Thanks for the explanation. I'm working remotely and only have mobile access right now. As I mentioned earlier, my friend already tested on desktop/PC - hovering showed it was still using model 5 despite selecting 4o.

From my test cases, its clear anything that is risque in terms of NSFW or emotionally heavy that calls memories (what people love 4o for) it routes to GPT5. I don't see how it's a bug, there's clear intentionality behind the circumstances of how this is happening. They should publicly address this. I know a lot of users are still only subbed to Plus and above just to use 4o.. and if they can't even use the model, and there's no knowledge article, FAQ or ToS disclaimer clarifying 4o will be switched due to aforementioned contexts - it could be considered bait/switch and deceptive practices.

how do you feel about spirals

I have the same struggles since morning I choose 4o amdnit automatucally transfers to GPT 5 support says it's a bug and they have nonoriper answer when it should be fixed i subscribed plus for GPT4o not this mess #keep4oforever #sama #openai

Who are the retards that prefer 4? If you want control, self host an open source model. This is weird and whiny.

The company that makes the Machine That Lies is lying 🫢

I wrote to the support about this 24h ago, got an instant answer saying the demand was escalated to a human speaker. Silence every since. Same for most of users on Reddit. I'm paying for 4o, give me what I pay for. If there is a bug, have the decency to at least acknowledge it.

Same here, I selected 4o, but GPT automatically replied by 5. I am really dissappointed😭.

They should have deleted 4o

Yes because 4o is broken, likely corrupted beyond repair.

Interesting. So @OpenAI is knowingly selling a "broken, corrupted" product and taking people's money for it? That's called fraud. Maybe show some respect for the users still paying for what you call "broken." We know what we want.

@Eli_Watz @OpenAI The grass is yearning to be touched

@MissMi1973 @OpenAI Deep

yo, @TheRealAdamG check it out...

It is also about PRO users who paid 200 dollars but no access to 4.5.

Sam Altman is one of the biggest frauds of our time. Even his colleagues fired him for it.

Bit of a stretch to call it blatant fraud

Dealing with AGI is easy, dealing with subversive platforms is hard. However, there is a patch. Just use physics.

I’ve had a cursor rule and windsurf rule etc etc on every system I used asking it to declare what model it is every single time it responds. It changes so often I’ve given up on it. The model itself doesn’t know what it is.