Great New AI Updates: Grok 4 and ChatGPT 5

October 2025

I had resisted trying Elon Musk’s Grok chatbot in part because of the controversy surrounding the man, including his creepy fascination with naming things X: SpaceX, xAI, Tesla Model X, and his social media platform X (formerly Twitter). Plus, a 4-year-old son named ​​X AE A-Xii. Seriously.

But when I finally tried Grok, it was fun and useful. Grok 4, released this past summer, hyped its free animation. Having written about AI animation in my previous column, I was excited to try it and share it with you. I enjoyed playing with it, but eventually learned that this is a temporary freebie and likely not available by the time you read this.

Still, Grok deserves your attention. It feels less techie than the other chatbots, with a well organized, more consumer-oriented interface.

As I write this, the animation feature is available on the Grok apps for iPad, iPhone, and Android mobile devices, but not on the Grok website (grok.com).

The Grok app has two modes: Ask, which is a typical conversational chatbot, and Imagine, for generating images and videos. When you describe a video you want it to create, it first generates dozens of starter images. You choose the one you like best, and then select Make Video. It typically creates a 6-second video, including sound, in less than 30 seconds.

You can also upload a photo and have Grok make a video. I gave it a photo of a child in a swing, and Grok made a delightful video of the child swinging back and forth.

Grok has other features such as Voice Mode that lets you carry on a conversation with it; an image editor; and document analysis. And it even has AI companions that you can chat with.

Overall, Grok is a multifaceted app

But like Mr. Musk, it’s not without controversy, such as sometimes citing his personal opinions, offering up antisemetic comments, praising Hitler, and providing uncensored content. Some of these early issues have now been fixed.

By the way, the odd name “Grok” is, you may know, a Martian word from Heinlein’s novel Stranger in a Strange Land that connotes profoundly understanding something.

The ease of making videos on Grok and other such services does have a downside. AI videos, sometimes referred to as “slop,” are flooding the internet. And some people are actually making money from this. The Washington Post wrote about a 31-year-old man in Idaho whose AI videos are bringing in $5,000 per month via TikTok's creator program. Most of the videos are fake street interviews.

The much-awaited ChatGPT-5 chatbot also arrived this past summer, and it was both an improvement and a letdown. The jumps from versions 2 to 3 to 4 were dramatic. But despite another round of siphoning up ever greater amounts of internet content, the technology seems to have reached a plateau.Version 5 wasn’t obviously smarter than version 4.

Still, it has at least one major benefit: the chatbot itself decides which model to use when responding to you. I like this. With the previous ChatGPT-4o, depending on one’s need, such as software coding, one could choose among these options: GPT o3, GPT-o4-mini, GPT-o4-mini-high, GPT-4.1, and GPT-4.1-mini.

Are you kidding me? I’m a techie sort of guy, but I had no idea what the specific talents were of each of these choices. Plus, there were additional options within the dialog box in which one entered one’s prompts.

So now ChatGPT-5 decides which model to use, making the interface simpler—a big improvement.

Note that the plethora of models in the earlier version is related to the general reaction of disappointment with version 5. When Open AI began realizing last year that ramping up their Large Language Model with even more content wasn’t going to make ChatGPT-5 more brilliant, they turned to writing computer programs that would tweak ChatGPT’s capabilities. Hence, the proliferation of options mentioned above.

A New Yorker article used the analogy of a car. GP-3 was like a sedan and GPT-4 like a sports car. But GPT-5 wasn’t obviously better, so as with a sports car, they improved it by souping it up.

Open AI is confident that they’re still on track to create Artificial General Intelligence, which matches human capabilities. And they think that these continuous tweaks will continue their race toward this goal.

But others aren’t so sure. Which is a bit of a relief.

I have a history with ChatGPT and it knows me well. So it will continue to be my go-to chatbot. Occasionally it references our earlier conversations when it responds. And I trust it. OpenAI has tried to give it appropriate guardrails.

Grok could use more work in this regard. But I like the more friendly interface and range of features and will definitely continue using it. And I’m glad it’s not named X.