Can the same thing be said about @janus's and similar experiences with Opus and other models? If no, how they are different?
>A Gemini subscription doesn’t give you access to a system prompt, but you should be using aistudio.google.com anyway, which is free.
As far as Gemini subscription (Gemini App) is concerned: you can create "Gems" there with a set of "instructions". Can using chats with such "Gems" be seen as an equivalent to adding a system prompt in ChatGPT / Claude?
Absolutely. Quality-wise the distinction is obvious. I guess my question was a bit vague. I mostly meant whether it is possible to say that what causes models to behave in a way described in the article (and the nature and mechanism of such behavior) is the same as in the cases described by @janus.