[Plus Subscriber] All models respond in dry, unformatted style after Jan 31 update. Free accounts un

[minor girder]

Reported by @obsidian helm

Bug Report: [Plus Subscriber] All models respond in dry, unformatted style after Jan 31 update. Free accounts un

`Steps to Reproduce`

1. Log in to ChatGPT Plus account on any device (iOS, Android, Web).
2. Start a new chat with any model (GPT-4o, 5.1, or 5.2).
3. Ask any question that previously triggered a detailed, formatted response (e.g., "Explain cat purring with enthusiasm").
4. Observe: response is dry, lacks emojis, bullet points, and emotional tone.
5. On the SAME device, log out and log in with a free account (no subscription).
6. Repeat steps 2-4. Observe: response is normal, formatted, with emojis and

`Expected Result`

As a Plus subscriber, I expect to receive detailed, formatted, and engaging responses from all available models, consistent with the service quality before January 31, 2026, and matching the output quality currently available to free accounts.

`Actual Result`

Since January 31, 2026, my Plus account receives only dry, plain-text responses from all models. All formatting (emojis, lists, dividers) and emotional warmth are missing. This significantly degrades the user experience and perceived value of the Plus subscription. The issue is account-specific and persists across all platforms and devices.

`Environment`

· Primary: iOS 18.3, ChatGPT app latest version from App Store. iPhone 16 Pro Max. · Also reproduced on: Android 14, ChatGPT app · Critical note: The bug is account-specific. The free account works perfectly in all the same environments.

Feb 1, 2026, 12:07 AM

Additional Information

Please provide relevant details to help resolve the issue, such as:

• ChatGPT Shared Link (if applicable).
• Screenshots or videos demonstrating the problem.

-# ➜ Need to contact support? Visit the OpenAI Help Center.

[obsidian helm]

Proof that Personalization settings ("More warm", "More emojis", etc.) are completely ignored for my account. The system prompt override is breaking core functionality.

[echo thorn]

Your base style and tone is set to “professional”. That takes precedence over the finetuning options and literally makes the models respond formal.
Maybe set it to friendly?

[obsidian helm]

No, changing the style doesn't help at all now. I've already tried everything. I'm waiting for a response from support

[severe needle]

"I'm waiting for a response from support"

Just so you know, that's not going to happen. Yes, this is the official server, but it still is a "community based" server.

There's only handful of actual OAI employees here, and none of those are going to respond.

So you're waiting on something that's not going to happen.

[obsidian helm]

Okay. But right now I'm communicating with support via email. I submitted a request at help.openai.com, and I hope they can help me. Have you ever encountered this? How is ChatGPT responding to you now?

[severe needle]

Oh yeah, via mail that's a different story. I thought you were talking about here on Discord.

But hm, I don't know tbh, I personally already have my GPT set so it doesn't use as much emojis, etc.

So I wish I could help you with that, I hope you get a response from them soon, but that might take a while.

I just put everything on default and it still doesn't use emojis. Still very "dry".

I'll keep it on default for a bit, and see if it continues.

[severe needle]

It definitely has something to do with the "professional" part that's causing "issues". I've just tested it.

I'll share a chat I just started.

GPT's first response was when I had everything on "more" and style on "professional"

I changed the style to "default" (still with everything else on "more") > responded to GPT and it gave me a response that's more aligned with the settings, but still somewhat "dry" compared to how you'd expect.

https://chatgpt.com/share/698087cf-3720-8011-a31b-aaabe5b59e3a