I have paid personal subs to ChatGPT, Claude, Grok, Gemini
For science (primary biology/pharmacology) questions, Gemini 3.1 Flash Extended produces the answers I _personally_ find "best", in terms of content, phrasing, and formatting.
For science (primary biology/pharmacology) questions, Gemini 3.1 Flash Extended produces the answers I _personally_ find "best", in terms of content, phrasing, and formatting.
However, I find the Gemini web app to be by far the worst, and Gemini itself second only to Claude in terms of refusing legitimate requests. It used to be the worst for that, but Claude has really put up the guardrails since their run in with the US president.
Grok has no concept of safety, which means that it can do certain things that none of the other models are allowed do, especially when it comes to research, creative tasks, humour and games.
Other things… well people are going to disagree on what’s “legitimate”, but perfectly legal queries about morally questionable topics are commonly refused. ChatGPT refuses to translate or OCR texts containing racial slurs, won’t help with information about building firearms, won’t give you information about how to stage a bank robbery, won’t talk about anything involving suicide, refuses questions around sex work and kink activities.
I can go to the library and rent a book on gunsmithing, sex work is legal, heist fiction is popular, racial slurs are unpleasant but common in certain texts. I totally understand why they want to block that stuff off, but I also don’t really like my computer telling me what is acceptable for me to think about based on someone else’s moral code. Also a world where games don’t include anything immoral or “unsafe” is probably quite bland.