They are battling in the leaderboards while Claude is doing real work.
In my experience Gemini 1.5 Pro was giving me more complete code than Claude, but the Gemini-generated code would not compile (C++ code) due to several errors. I was going back and forth with Gemini and it became clear that it was in a "loop" (it would suggest a fix and then that wouldn't work and it would suggest the original code again, wash, rinse, repeat). Then when I asked Claude why that code wouldn't compile it immediately identified the problems (mainly due to libclang API changes over various versions) and the code compiled.
it doesn't matter than Gemini is crushing the leaderboards because ChatGPT has the fanboys
it's basically like Apple at this point, they don't have to be the best
I think it's the exact opposite honestly. I use chatgtp because it's not a google or facebook product and this isn't a unique sentiment among my developer friends.
The linked article showed the opposite for everything except coding though..?