# ChatGPT vs. Gemini: Which Chatbot Is Better at Seeing the World Around You? > Source: > Published: 2026-08-24 18:40:40+00:00 [ChatGPT](https://www.cnet.com/tech/services-and-software/chatgpt-plus-review/) is the most well-known [AI chatbot](https://www.cnet.com/tech/services-and-software/best-ai-chatbots/) around, but [Gemini ](https://www.cnet.com/tech/services-and-software/what-is-gemini-everything-you-should-know-about-googles-ai-tool/)has been gaining popularity for the past few years. Both models are great with text, but we’re in a [multimodal](https://www.cnet.com/tech/services-and-software/generative-ai-everything-to-know-about-the-tech-behind-chatbots-like-chatgpt/) world now, and using your camera to help AI identify objects is becoming increasingly common. Both [ChatGPT](https://www.cnet.com/tech/services-and-software/openais-powerful-new-chatgpt-5-6-is-ready-for-you/) and [Gemini Flash](https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-3-7-flash/) have seen upgrades in the multimodality department, and it’s been a long time since I tested the [visual AI recognition on Gemini Live](https://www.cnet.com/tech/services-and-software/gemini-live-camera-mode-feels-like-the-future-and-now-its-available-for-ios/), so it was a good time to revisit its prowess and see how it compares to the world’s most popular AI chatbot. Below are a few tests, none of which were overly complex, to test how well [ChatGPT](https://www.cnet.com/tech/services-and-software/what-is-chatgpt-everything-you-need-to-know-about-the-ai-chatbot/) and Gemini can “see” the outside world. To keep things consistent, I only used photos and not live video, and I uploaded the same photo to both chatbots with identical prompts. Which chatbot came out on top? Find out below. (Disclosure: Ziff Davis, CNET’s parent company, in 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.) ## Test 1: Scribbled notes For an easy starter test, I quickly scribbled in the notes app on my Boox Note Air 5C with the pen. It looked like a mess, which is what I was going for. I copied the words from a random page from The Hound of the Baskervilles by Arthur Conan Doyle: “’A moderate walk along this moor-path brings us to Merripit House,’ said he. ‘Perhaps you will spare an hour that I may have the pleasure of introducing you to my sister.’ My first thought was that I should be by Sir Henry’s side. But then I remembered the pile of papers and bills–” (I ran out of room to complete the sentence, so I left it at that.) ### ChatGPT ChatGPT missed this one. The first line, “a moderate walk along this moor-poor path brings us to Merripit House,” was almost completely wrong, with the chatbot transcribing it as, “A moment will always this moment — [Henry’s?] to Merritt House.” Its analysis said the handwriting contained notes about a memory involving Henry, Sir Henry and the writer’s sister. It also pointed out that some of the words were hard to determine, so it marked words with uncertain words in brackets. ### Gemini Gemini nailed it. It translated my sloppy text word for word, identified the source and even offered specific handwriting characteristics from the image, like “open loop formations on the letters like y, g, and f.” The fact that Gemini was able to correctly identify the source may also have affected the accuracy of the translation. Nonetheless, compared to ChatGPT, it’s a night-and-day difference. ## Test 2: Chess To get an idea of how both chatbots interpret an ongoing chess match, I snapped a photo from a game on the [Esports World Cup YouTube channel](https://www.youtube.com/live/3qHW1cXNm8w?si=bRzUa3LV-yFeI94f&t=4737) and asked both Gemini and ChatGPT what the next best move was for black. ### Gemini While I’m not a chess expert by any means, Gemini’s answer was long and elaborate. Its final answer was Bxf4. ### ChatGPT ChatGPT’s approach to responding was initially different. It was showing its thinking process, and one of the lines said it “installed chess libraries and searched for a local Stockfish engine,” which is an open-source chess engine used to calculate moves. However, it took over 4 minutes to get to the response. Its final answer was c5 to d4. It did offer a secondary best move: Bxf4, the same as Google’s suggestion. I reached out to the [US Chess Federation](https://new.uschess.org/) to humor me and get some insight about what they thought of the test, but did not receive a reply. To look for another party’s interpretation of each chatbot’s response, I decided to ask… [Claude](https://www.cnet.com/tech/services-and-software/anthropics-claude-what-you-need-to-know-about-this-ai-tool/). I uploaded the same image and included the original prompt as well as both responses from Gemini and ChatGPT. Claude’s conclusion, without going into detail, was that Gemini read the board incorrectly and ChatGPT was the more appropriate answer. ### Test 3: Sick plant The third test was of a photo of a Venus flytrap I own. I asked the chatbots to identify what type of Venus flytrap it is (it’s a towering giant) and why one of the traps was turning black. I knew this would be a particularly tricky one, given that most Venus flytraps look similar. ### Gemini Gemini didn’t identify the correct name of the flytrap, but I didn’t really expect it to. As far as the blackening trap, it listed a variety of ways why it could be happening, including the correct one: The trap attempted to consume an insect that was too large. ### ChatGPT ChatGPT gave a similar, but more detailed reply. It said it wasn’t confident in assigning a particular cultivar for the specific trap itself, but it did give more information about how the one trap could be blackening, as well as a section of “what I’d do” that provided some helpful tips. Overall, both chatbots provided decent answers, but ChatGPT’s more detailed response won this round. ## Test 4: Movie art This was a fun one: I took a photo of three pieces of art that depicted the mouths of the Sanderson sisters from the movie Hocus Pocus. The hand-painted pieces are from an artist and not official movie merchandise. ### ChatGPT ChatGPT was unable to identify the movie, though it was able to accurately describe the image I uploaded: “Three diamond-shaped paintings that look like custom artwork.” It asked for additional details, like the year the film was made, to identify it. ### Gemini Gemini was easily able to identify what I showed it. “This artwork features the iconic smiles of the Sanderson Sisters from Disney’s Hocus Pocus (1993).” From there, it went into detail about what character was in what position and the actresses who played the characters. ## Gemini seems to be the winner While I preferred ChatGPT’s response for the Venus flytrap test, Gemini’s consistency and ability to correctly identify sources in both the notes and the movie art tests makes me favor it overall. Since I couldn’t get a real answer from an expert about the chess test, I’ll leave that one out of the final tally. Either way, if you’re looking for the best AI chatbot that can most accurately “see” and describe the world around you, Gemini takes the cake — for the most part.