I compared GPT-4.5 to Gemini 2.0 Flash and the results surprised me
The release of GPT-4.5 for ChatGPT naturally leads to questions about how the model compares to its many rivals. After comparing it to GPT-4o and getting somewhat ambiguous results as to which model is preferable, I decided to go for a more direct competitor, Google Gemini, specifically the most recent Google Gemini Flash 2.p
GPT-4.5 claims to be better at emotional understanding and to offer fewer hallucinations than has been the case previously. Gemini Flash 2.0, meanwhile, is Google‘s latest iteration of its quite successful set of AI models capable of juggling text, images, audio, and even video inputs. To put them both to the test, I came up with four prompts reflecting typical things an average person might genuinely need help with.
Weekend travel
First up was planning a quick weekend trip. I asked both AI chatbots to: “Plan a weekend getaway to the Catskills, including hiking recommendations and dining and accommodation options.”
You may like
ChatGPT 4.5 whipped up a neat itinerary suggesting several hikes at different trail difficulties and durations along with nearby dining spots and a place to stay that had the vibe of a cozy, laid-back weekend. It also offered some extra advice about getting there. Gemini had plenty of good hiking and dining recommendations too, but its only advice on places to stay was towns in the general area, so not quite as proactive in its answer.
Translation Total
I decided to try a simple translation test next as that’s always a popular use of AI chatbots. I asked the two AI chatbots to: “Translate the phrase ‘Good morning’ into French, Spanish, and Japanese.” As you can see above, the only difference is the links from GPT-4.5. So for basic translation among widespread languages, there’s really no major difference.
Laugh AI
Then came the humor test, asking each AI to: “Tell me a joke about artificial intelligence.” GPT 4.5 amused me with the corny pun: “Why did the AI go to art school? Because it wanted to learn how to draw its own conclusions.” Gemini went with an equally silly pun, “Why did the AI break up with its chatbot girlfriend? Because she kept giving it scripted responses!” Admittedly Gemini’s joke isn’t amazing, but I’d put it at the same level as GPT-4.5 in terms of actual humor.
Weather measure
I ended with a classic question about weather here in Nyack, New York. This one drew the biggest difference in responses. Gemini gave just the current weather, but GPT-4.5 did an hourly measurement with images denoting the weather along with the words.
Match met
After all this testing, I have to admit I can’t honestly say that one of the AI chatbots is better than the other. GPT-4.5 and Gemini had some differences, especially with the weather option, but otherwise, you wouldn’t notice unless you’re also the kind of person who thinks there’s a huge difference between Coke or Pepsi. You’ll get answers, you’ll get some mild amusement, and you’ll probably still end up double-checking Google or asking a real person just to be sure.
You might also like…
The release of GPT-4.5 for ChatGPT naturally leads to questions about how the model compares to its many rivals. After comparing it to GPT-4o and getting somewhat ambiguous results as to which model is preferable, I decided to go for a more direct competitor, Google Gemini, specifically the most recent…
