This game was recommended because I was looking for a game that uses an LLM, but with a predefined story and more restrictions than just using a character card. I want to run this game locally, but I read here that some don't like the responses. Is the LLM used not good? Are there any recommendations what works better? I have an RTX 4090 with 24GB VRAM, so that would be the max I can use. My current go to model is EVA-Qwen2.5-32B-v0.2_EXL2_4.1bpw_H8 with SillyTavern and I am more or less happy with most of the responses. Could I use that model instead? Or would it be too big, as it alone is 17.0GB big. And I don't know how much Multiic needs for context, tts, image generation or whatever it else includes.
So maybe a general recommendation of how big it max could be would be better? If there isn't already a model most of you think is the best to choose.