Hapabapa/Getty Images

Artificial Intelligence (AI) has a long and complex history, and is something that tech giants and research labs have been working on for decades now. However, it only truly caught the public’s attention when OpenAI launched ChatGPT in November 2022 and crossed the 1-million-user threshold within just five days.

Fast forward to today, and that figure has grown exponentially. It’s fair to assume that ChatGPT is now the first tool that comes to the average user’s mind when they hear the word “AI.” And much like its ever-growing user base, ChatGPT shows no signs of slowing down. OpenAI continues to iterate at a relentless pace by launching models and new features left and right, and 2026 might just be shaping up to be its most transformative year yet. 

Here are a couple of big changes coming to ChatGPT this year that you should know about. These come hot on the heels of new personality presets and ChatGPT Health, which launched in late 2025 and early January 2026, respectively. Keep in mind that the information below is based solely on what OpenAI has chosen to reveal publicly. If history is any indication, there’s likely much more in the pipeline that we don’t know about. And if OpenAI is listening, here’s hoping that includes a few features ChatGPT still desperately needs in 2026.

subh_naskar/Shutterstock

If you’ve been following the AI world closely, you likely know that companies like OpenAI drop new models every few months. Rather than mere fancy version numbers and minor speed bumps, each new model is built to be smarter, faster, and more capable than its predecessors. Shortly after a new model is launched, older ones are typically phased out and retired.

On February 13, 2026, OpenAI will be retiring six models in one go. The AI giant will discontinue the following models: GPT-5 (Instant and Thinking), GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini. These models will be removed from ChatGPT, though they’ll remain available through OpenAI’s API. From that date onward, GPT-5.2 will be the default model for all ChatGPT users.

Interestingly, this is the second time GPT-4o will be deprecated. OpenAI first attempted to retire it during the GPT-5 rollout but had no choice but to restore access after Pro and Plus users shared that they preferred GPT-4o’s “conversational style and warmth.” Now, OpenAI says it has taken that feedback into account and addressed its shortcomings. The company claims that the vast majority of users have since switched to GPT-5.2, with just 0.1% of its user base still using GPT-4o.

Yarrrrrbright/Shutterstock

No one likes ads. In October 2024, OpenAI CEO Sam Altman publicly said that he disliked ads and the prospect of mixing ads and AI. Altman went so far as to say that serving ads to users would be a “last resort” for the company. A mere 20 months later, that last resort has arrived. In mid-January 2026, OpenAI confirmed plans to begin testing ads in the United States via its social channels and a press release.

The company announced this alongside news of the “ChatGPT Go” tier coming to the U.S., a new $8/month subscription that sits between the free plan and the $20/month Plus subscription. OpenAI explained that both free users and Go subscribers will see ads, while those on Plus, Pro, Business, and Enterprise will continue to enjoy an ad-free experience. The company emphasized that ChatGPT’s responses would never be influenced by advertising and that any ads appearing within ChatGPT’s interface will always be separate and clearly labeled.

Additionally, OpenAI claimed that users’ data and conversations are protected and will never be sold to advertisers. Interestingly, the company reiterated that it prioritizes user trust and user experience over revenue, even though the introduction of ads feels like the antithesis of that claim.

To soften the blow, OpenAI has added that ads won’t appear near sensitive or regulated topics like mental health, healthcare, and politics. Currently, ads are being tested in the United States for logged-in adult users. So, if you haven’t seen any ads pop up in ChatGPT and are on the free or Go tier, consider yourself lucky… for now.

Thampapon/Shutterstock

Now, young children having easy access to the internet was scary enough. Add AI chatbots that can answer practically any question without needing much context, and the concerns multiply fast. Fortunately, this is something OpenAI isn’t taking lightly. The company has started rolling out an “age prediction” feature globally.

As the name implies, ChatGPT will attempt to predict a user’s age and determine whether they are over or under 18. In a support article, OpenAI explains that it uses an age prediction system that looks at different signals linked to a user’s account. This includes the general topics they talk about and even the times of day they use ChatGPT. 

For example, if a user is consistently active during after-school hours or late at night on weekdays (times typically associated with a younger demographic). ChatGPT might flag the account as potentially belonging to a minor. Similarly, if they frequently discuss topics common among younger audiences, ChatGPT may classify them as under 18.

If the system determines that someone is under 18, extra safety settings will be automatically enabled. As a result, sensitive topics like graphic violence, sexual role-play, viral challenges, self-harm, and unhealthy body image content will all be handled with extra care. Of course, there may be times when the system incorrectly flags someone as under 18. In such cases, users will have the option to verify their age manually on the ChatGPT website.

Saeidreza maghsoudi/Shutterstock

AI responses, including ChatGPT’s, have always been… boring. How many times have you begged it to spare you the long walls of text and just get to the point? Fortunately, this is something the folks at OpenAI have clearly heard. In fact, they’ve already done something about it, although the exact details are still a bit fuzzy.

On January 30, 2026, OpenAI updated ChatGPT’s release notes with a somewhat vague announcement, claiming that responses are now “more visual and easy to scan.” According to the notes, answers to everyday questions may include “at-a-glance visuals,” while important people, places, products, and ideas will be highlighted in-line. Tapping one of these highlights opens a side panel with key facts and trusted sources.

Unfortunately, this is all the information the company has provided at this time. While there isn’t much clarity on what this will look like in practice, early reports suggest it’s similar to Google’s knowledge panels. That said, the feature has begun rolling out globally across ChatGPT on the web, as well as the Android and iOS apps.