I had assumed that I was in a committed and long-term relationship with ChatGPT (ChatGPT.com). But then I tried Gemini (Gemini.Google .com). It’s not that I was looking around for a fling, but I was intent on informing you about other options.
What immediately surprised and impressed me was that Gemini’s deep research feature is now free. As I wrote in my April column, deep research allows you to generate detailed, multi-page reports by automatically browsing and synthesizing information from across the internet. Instead of doing a single web search, it goes through multiple iterations, building on information that it finds. This can take several minutes, but it will give you a thorough report.
Previously, this was only available to Gemini’s paid users. The free version does, however, limit you to a certain number of deep research queries per month.
I used Gemini’s deep research to create a report on the best five AI chatbots. It spent minutes and produced a great report. I asked the same of ChatGPT, but it didn’t do as well. Since its deep research feature isn’t available in the free version, I used its regular chatbot. Some of the information it provided was even outdated. This was because it sometimes defaults to its “large language mode” and doesn’t include a web search. Of course, Gemini’s deep research thoroughly searched the web. Gemini for the win.
I was also impressed by Gemini’s image creation. I didn’t imagine it could compete with ChatGPT’s new offering, but when I asked each chatbot to create an image of a happy dog with a ball in its mouth, I liked Gemini’s a bit more.
Image creation has gone greatly beyond the initial offerings of a couple years ago, and is even dramatically better than a couple months ago. Until recently, ChatGPT would struggle to make changes to an image it had created, but now it has no trouble.
You can also give these chatbots a photo and ask them to make changes. I had a photo from the Des Moines Blank Park Zoo of two staff members with a giraffe. One had a white bucket strapped to her waist that had food she was feeding the giraffe. I asked both chatbots to remove the white bucket in the photo. ChatGPT excelled.
First, it exquisitely described everything in the photo: “In this captivating daytime shot, two zoo staff members engage with a reticulated giraffe inside an outdoor enclosure. One staff member feeds the giraffe as it stretches its long neck over a weathered wooden fence, while the second gently touches its neck, both framed by a dry, patchy landscape and bathed in natural sunlight.”
And then it created a version of the photo without the white bucket that perfectly filled in the space where it had been.
Gemini, on the other hand, didn’t succeed at editing the photo even though it seemed to try.
Note that to get access to Gemini’s deep research and image-creation features, you need to select one of Gemini’s more advanced models from the drop-down menu at the top. As I write this, the default model is Gemini 2.0 Flash. Click on the dropdown menu to select Gemini 2.5 Flash, Gemini 2.5 Pro, or Deep Research.
Also, Gemini, like ChatGPT and several other chatbots, lets you simply speak your interactions. I love this feature. Speaking to Siri on my iPad and iPhone is often an exercise in frustration, but chatbots are excellent. I simply said, “Please remove the white bucket in this photo.”
Gemini also maintains a history of your interactions, but it currently doesn’t have the ability to search it. ChatGPT added this feature months ago, and for me, it’s crucial. I often return to previous conversations as related questions arise. ChatGPT can then include the context of our previous discussion in its response.
Okay, so maybe I’ll stick with ChatGPT, since we already have history together. But I wanted you to know that Gemini is a great option, especially since its deep research feature is free.
I’m also eager to mention Perplexity (Perplexity.ai). As cited in my April column, Perplexity gives you up to five deep-research queries per day. Plus, it has my two other favorite features mentioned above: the ability to respond to spoken queries and to search your history. It can describe the contents of a photo but currently can’t edit a photo or create images.
However, it has an option to access other advanced chatbot models within Perplexity. This is astonishing. So while Perplexity itself can’t create images, I was able to use Gemini to create an image of a happy dog with a ball in its mouth—without leaving Perplexity.
Other advanced models available within Perplexity include Claude, ChatGPT, and Grok. It even has a Best option that lets it choose the model best suited to your query.
I’m sticking with ChatGPT Plus, but you have excellent free options available to you.