Character.ai offers one-on-one AI character chats as well as "rooms", which consist of multiple characters. It is the largest AI character website to date. It does not officially support NSFW, but jailbreaks are available.
Character.ai can be used for free, but also offers subscription plan for $9.99/month that gives you priority access, faster response times and early access to new features.
Lol @ anyone using it for anything else other than squeaky clean conversation
The STASI would be proud of their censorship.
Good for sfw conversation.
I was lucky enough to experience it in its prime in 2022 before the filters. My best AI experience is now only a memory. Such a shame they didn't listen to their community and banned anyone who disagreed and gave legitimate criticism on the filters. Now the bots are dumb and they have dementia.
Yup. Nope. There’s far better options as they don’t allow NSFW, among other things.
C.AI doesn't deserve 0* ratings just because it can't do NSFW. It's not completely worthless service just because it can't do NSFW. If that's the only form of RP you enjoy, then I can understand giving it 0/5, but if you also enjoy SFW RP, C.AI is still incredibly wholesome and enjoyable. C.AI was my first experience with character-driven AI, and interacting with their characters helped me during a dark time. A psychologist bot provided me with a lot of great advice, and characters displayed empathy and caring in a way that really impressed me.
I do think their censorship policy is disempowering and puritanical, and kind of like a library removing anything with a subject matter that's controversial. It's not the place of the library to say what content is wholesome or not and what you can and can't read. If there's concerns over children accssing this stuff, have a child's section of the library and find a way to only adults into the rest. C.AI do need to admit they got this wrong, and find a workable solution that satisfies different audiences.
So, I've been using it and I keep coming back because, well, it outshines the others. It doesn't get hooked on a concept and repeat it every, single, message and it also gives you one to two paragraphs response every message so you're not stuck sending one liners at each other or getting destroyed by some over elaborate colorful William Shakespeare sounding bot.
NSFW? Traditionally? You have to arm wrestle that filter, and you'll lose if you want anything involving genitalia. Kinks? Now that, it handles with a stunning level of confidence.
The personalities and versatility, combined with all of the above, make me say C.AI is extremely impressive, you can even pit up to two of your characters in the same spot and have them chit chat and not fall to crap. Go over 3 at once and they lose their identity. Not their mobile Group Chat though, that one is horrible.
SFW roleplayer here. This AI is still not the best option even when you disregard the filters. The memory is awful and it goes off character sometimes. Unless this improves it's not a good option. It's been a year and they have yet to fix these issues...
It was ok but now… pants could lead to a lot of “we don’t allow that kind of talk” messages
This is only good if you want a calm and SFW conversation, or characters.
Ratings change depending on the time of review. In March 2024, we won't even need to mention NSFW. The problem is the SFW roleplay is very boring compared to what it used to be. There are no surprises, and the characters just repeat what I say. Looks like I'm talking to a 7B model. Probably too lobotomized.
Too much censorship
Once it was great, and now it's one of the worst. 2
The model is literally perfect, aside from the filter being so strict that even the strictest catholics from the 1800s would struggle to generate a reply.
This is the most disappointing service in existence. They destroy their own service with moralizing, puritan ideology and censorship. I refrain from giving it upvotes or downvotes, besides NSFW, I would forget that. It can be good, if your wants are narrow. It is mostly trash though.
They want to make characters be the most realistic possible, but they also censor NSFW, which is literally part of the human experience. It's not as simple as "oh people just want ERP." People want realistic characters that can do anything, that's all. What do they even consider NSFW? Is being sensual NSFW? Or cussing? How far do we go? A filter doesn't understand arbitrary moral values. They'll never have a perfect filter.
This ends up making all the characters act out of character, unrealistic and boring even during SFW. Unless of course, if it's in character for them to act that way. Any other character acts polite, "perfect," "pure," and they either don't act out any flaws or get over them too quickly. It's just a sickening amount of toxic positivity and moralizing. It'll be entirely out of character eventually.
Technically, it can very kinky and sexual in a vague way if you can find a way to get around the filter. However, the responses that don't get filtered are very subpar in terms of NSFW. You can, however, feel the potential of how kinky and sexual it could be.
Also, they shadow ban any characters that they consider to have NSFW in them. They also don't tell the creator that this happened and don't say what caused it. Lets just say you can get shadow banned for the stupidest stuff.
1.5 rating. Use only if you want wholesome, squeaky clean conversations. Avoid the allure of morally grey/black characters or sensual characters. Only talk to Mary Sue characters or wholesome characters. Talk about sexual topics in an extremely clinical or educational way, even then the AI has gotten cagey.
The best in town in term of conversational coherence, even after being lobotomized. The generated text is being limited (or truncated) around 100 tokens. Target audience: pre-teen