As someone who spends a lot of time evaluating AI chatbots, I find the launch of GPT-5.1 as disappointing as that of GPT-5. Despite OpenAI trumpeting advances in response quality, a warmer, friendlier tone, and ways to make ChatGPT uniquely yours, the latest update largely doesn’t impress me in practice. I encountered flashes of a fresh personality, but nothing that drastically changes the chatbot’s feel. Here’s what you should (and shouldn’t) expect from the newest iteration of ChatGPT.
What’s New With GPT-5.1?
According to OpenAI’s blog post, GPT-5.1 (which is out now) is faster and more intelligent than GPT-5 across the board. This might be true in certain scenarios, but I haven’t noticed a significant difference in GPT-5.1’s response quality or speed in real-world testing. That said, GPT-5 (and even GPT-4o) didn’t exactly struggle in those areas in the first place.
Beyond performance enhancements, OpenAI says that GPT-5.1 is “warmer and more empathetic” in general. This is a somewhat stark reversal from earlier in 2025 when OpenAI updated GPT-4o’s personality to be less friendly and sycophantic. OpenAI also added several new personalities to ChatGPT that further impact the model’s tone. GPT-5 had Cynic, Listener, Nerd, and Robot personalities, but GPT-5.1 changes Listener to Friendly and Robot to Efficient, while adding the Candid, Professional, and Quirky options.
It was already possible to use custom instructions to dictate exactly how you wanted ChatGPT to behave, but GPT-5.1 supposedly adheres to those instructions better. This dovetails nicely with GPT-5.1’s personality and tone changes, because you’re free to adjust things if the default choices don’t quite appeal to you.
GPT-5.1’s Tone Shifts Are Very Subtle
Before I talk about my impressions, it’s important to keep in mind that my testing isn’t definitive. Your experience with ChatGPT might vary significantly due to the variable nature of chatbots. However, since I use ChatGPT regularly and am very familiar with how it behaves, I was able to pick out some occasional and slight differences between how GPT-5.1 and GPT-5 respond to prompts.
For example, when I asked both models about what one can do to be less anxious, GPT-5.1’s responses did feel warmer and more empathetic than GPT-5’s. But this wasn’t the case when I asked about farming loot in the video game ARC Raiders. So, yes, you can expect a slightly kinder ChatGPT with GPT-5.1, but only if your prompt opens the door to a more personable response. And again, it’s highly dependent on the topic.
Get Our Best Stories!
Your Daily Dose of Our Top Tech News
Sign up for our What’s New Now newsletter to receive the latest news, best new products, and expert advice from the editors of PCMag.
Sign up for our What’s New Now newsletter to receive the latest news, best new products, and expert advice from the editors of PCMag.
By clicking Sign Me Up, you confirm you are 16+ and agree to our Terms of Use and Privacy
Policy.
Thanks for signing up!
Your subscription has been confirmed. Keep an eye on your inbox!
Personalities and Custom Instructions Make for Mild Changes
It’s a similar story with GPT-5.1’s new personalities. For example, if you say “heyo” to GPT-5.1 with its Quirky personality engaged, ChatGPT responds, “A small hello rolls in like a marble across a tabletop.🎱What’s the vibe today?” If I do the same thing with its Efficient personality, it says, “Hey.” A personality shines through only when your prompts give it the space to do so. A generic greeting has that potential, but a technical question doesn’t. Nonetheless, if you want a quirkier ChatGPT, feel free to use the corresponding personality.
What if you don’t want to use a personality preset but instead want the chatbot to take on a more specific persona? In that case, you can give ChatGPT custom instructions, which OpenAI says GPT-5.1 should be better at following. I told GPT-5.1 to “Speak only like a surfer dudebro from the late 1990s living outside the Santa Monica beach would.” Accordingly, GPT-5.1 peppered “bro” and “ride the wave” and “surf’s up” into responses, but so did GPT-5 when I gave it the same instructions.
It’s entirely possible that GPT-5.1 adheres to instructions better than GPT-5 over the course of longer conversations, but I didn’t notice a major difference in my casual chats. Overall, OpenAI’s somewhat anemic implementation of personality across GPT-5.1 and GPT-5 strikes me as disappointing. Adding in a few new words here and there doesn’t make a world of difference, and you generally need open-ended prompts to experience the effect of custom instructions or a personality preset in a major way.
Recommended by Our Editors
GPT-5.1 Is Worth Trying, But Don’t Expect Too Much
If you felt GPT-5’s personality was a downgrade from GPT-4o’s, you might see GPT-5.1’s custom instructions, new personalities, and tone changes as moves in the right direction. That said, you might still be disappointed since the changes aren’t radical in most instances, even with custom instructions.
If you didn’t care that much about ChatGPT’s tone in the first place, the good news is that GPT-5.1 doesn’t feel quite as sycophantic as GPT-4o’s did at times. You can also just safely ignore its new personalities and not bother with custom instructions. I don’t like an especially friendly (or quirky) chatbot or one that’s overly cold and robotic, so I’m happy to use ChatGPT’s default setting. I largely don’t notice much of a difference with the new model based on what I usually ask the chatbot, and your experience is likely to be the same if you didn’t click with GPT-4o’s behavioral tendencies.
If you really want an AI companion, I recommend using a dedicated service. AI chatbots, be it ChatGPT or even Grok with its virtual characters, simply can’t match the robust personality customization options that AI companion apps specialize in, even if you can push chatbots to adopt slightly different tones and behave in slightly different ways.
About Our Expert
Ruben Circelli
Writer, Software
Experience
I’ve been writing about consumer technology and video games for over a decade at a variety of publications, including Destructoid, GamesRadar+, Lifewire, PCGamesN, Trusted Reviews, and What Hi-Fi?, among many others. At PCMag, I review AI and productivity software—everything from chatbots to to-do list apps. In my free time, I’m likely cooking something, playing a game, or tinkering with my computer.
Read Full Bio
