AetherRoom is a new chatbot service that will be coming out sometime in 2025 or the later half of 2024, (not exactly sure).
AetherRoom is created by ‘Anlatan’, The same people who made NovelAI… AetherRoom will be an 18+ competitor to Character.AI, Infact it will probably push past most of the competition because it is created By Anlatan… They make great AI tools that just work™
Anlatan will probably be a paid subscription service that will cost around 10 dollars a month at it’s lowest tier Just like NovelAI… So relatively cheap.
Some links for you to look over.
https://aetherroom.ai - You can apply to the closed alpha here
https://www.youtube.com/watch?v=8pISd87_EPI -Latest Devblog
1 Like
Is it an advertisement post or something? Because “they make great AI tools that just work” is absolutely not true, at least in my experience. I’ve always found NovelAI to be pretty awful at writing stories
2 Likes
I personally used novelai, and its great for writing stories… in my experience.
I’ve got half an eye on Aether room. I’ve dabbled quite a bit in Novel.ai, and it’s not bad, but not great either. It’s capable of some really good stuff, but even on the most expensive tier (yeah, I’ve spent money on it), it’s very inconsistent and forgetful on core memories and basic stuff. Most of the updates recently have been in the image field - but it only does furry stuf and anime style, neither of which I’m interested in. There are far better image generators out there.
Will probably try Aether room when it comes out, but it’ll have to improve a lot on the current Novel.ai experience to stand out, IMO.
I dont have a problem with the memory, I try to keep all my stories neat and tidy by putting lorebooks for almost everything that I can think of.
I pay for NovelAI, and will consider AetherRoom when it comes out. My experience with NovelAI is that it’s “pretty okay,” which is basically what I’m looking for. I use it mostly as a writing assistant/GM assistant, and it works rather well for that purpose - I rewrite probably 85% of the output from the AI, but that keeps it in my “voice” and I’m primarily looking for inspiration versus something to actually do the work for me. I’ve toyed with the image generation as well, but don’t commonly use it because I don’t want to include pure AI-generated stuff directly in things I make and I don’t have the art skill to use the outputs as a “outline to develop from” in the same way I use the written output.
From what I’ve seen? There are better generators for most image generation, but that’s because there are thousands of those and you can find one tuned specifically to your needs. Similarly, you can “tune” your own LLM instance if you want for a more tailored experience which will closer-match your desired outputs. In both cases, however, you’ll need to run it locally or pay for the processing from somewhere. It’s also a process, finding the right model and tuning it to suit the output you’re looking for. Comparatively, NovelAI gives a “good enough” result out-of-the-box with minimal fussing and has already-established good documentation on how to phrase inputs to get what you’re looking for.
I look at it as the “buying a Dell PC” of AI generation. Yes, you can custom-make something if you want better performance for specific purposes. But this generally works for most use-cases that most people will have in a good enough fashion that doesn’t require fiddling and time investment in making something yourself.
Whenever AetherRoom launches, I’ll probably give it a shot if there’s some sort of “bulk deal” for subscribing to both or if it gets lots of good reviews. If so, I’ll let you know what it’s like.