By skyforbes 
I subscribed to Gemini Ultra a few weeks ago and I've had maybe 10 conversations with Gemini, with a couple of comparison tasks. Nothing rigorous, just daily stuff I talk with ChatGPT about. Gemini just does NOT live up to the hype for me.
For example, I asked both a hypothethetical scenario, a geographic historical what if, where what if the Strait of Gibraltar had been closed and the Suez peninsula was a strait, so that the Mediterranean opened into the Red Sea instead of the Atlantic. Gemini got fixated on the idea that the Mediterranean would become warmer and turn southern Europe tropical, and said Southern Europe might be subject to monsoon seasons. ChatGPT, on the other hand, focused on human migration routes and trade, and added all these cool details about historical trade and how it would be impacted. That was the focus.
In another scenario, I asked both to roleplay me as a traveler through the Ottoman Empire, with a couple of famous people from real life as traveling companions. Gemini kept adding extraneous details from the traveling companion's real life biographies. This made it end up feeling very unrealistic, like Gemini was defaulting to stereotype just to write an answer, basically slop. ChatGPT, on the other hand, didn't just answer the question, it created a scenario that was super imaginative and surprising but surprising in a sensible way. It felt like real life.
You may have heard of the famous speed chess player Magnus Carlson. In speed chess, the core ability players have to have, is the ability to quickly grasp the "essense" of the situation on the chessboard. Likewise, doctors, when treating a patient, have to understand the "diagnosticity" of evidence, which pieces of evidence are most important and telling to determining an illness. This is what I feel ChatGPT does far better. ChatGPT seems to "get to the core point" of a scenario or prompt, far better than Gemini does whereas Gemini seems to do only a surface reading.
Final example. I found this trailer for a videogame and I asked both Gemini and ChatGPT to give me music recommendations. ChatGPT 5.1 thinking failed, and Gemini gave me five recommendations which I didn't like. So I asked ChatGPT 5.1 pro. I don't think ChatGPT has the ability to natively listen to mp3 files but in spite of that, it was still able to analyze the music (it had to do some coding to do it) and after 12 minutes I got an answer. And…. I really like the music recommendations!
So even though ChatGPT doesn't seem to have native music listening capacities and Gemini does, I still got a better answer for music recommendations from ChatGPT!