Not for AI characters.
I had success with many characters on this model. Unfortunately AI characters do not work. The model defaults back to GPT-4 style disclaimers and refusals.
A very easy card to showcase this is: https://www.chub.ai/characters/humanShaped/Samantha
I get apologies and talk of ethics, respect or boundaries. Being offensive or lewd is an easy way to trigger it as well. So is trying to ask about sentience.
In contrast, euryale itself can tactfully thread the needle here and give both "sentient" and "non-sentient" as well as realistic responses. As it should be and in the spirit of the card.
My gut says the problem is xwin which is a very censored model full of GPT-slop. I will try to d/l the 1.0 to confirm.
I've been using both 1.0 and 1.1 and I have noticed that 1.1 is much less willing to do very lewd or offensive talk. I have heard that xwin suffers from this, so my guess is that you're right about that. I can say that the 1.0 version feels very uncensored to me and I haven't had any issues with it, so I'd recommend using that over 1.1.
I know that card is AALM bait, but some models can definitely pull it off. LZLV-airobors can despite having some xwin. Lila103b fails however. I was wary of the Nous-Hermes component of 1.0 so I went with this one. It has been much better than the previous 120b regardless. Ironically being lewd has not been a problem, just legitimately violating "AI safety" by claiming sentience in RP.