How is the self-hosted AI?

[のつもALL]

Is this good enough to play the game, or is it too dumb and needs higher-end AI integration?
I still haven’t seen any videos of this setup. Does anyone have a video showing gameplay with self-hosted AI?

[sgthale]

I’ll make a video

[BlackusPL]

Same question because in old builds we had dropdown list where we could choose self-host, patreon, player2, etc. Now I don't see this option
EDIT: I found a way

[Balmung]

It of course depends on your hardware. If you are lucky enough to have a 24GB VRAM graphic card or a second computer to run the AI stuff on it, you still can have a very smart AI robot, since simple talk didn't need such a smart model as if you do real roleplay with longer dialogs. But even 16GB VRAM could be enough, but it also depends if you also want to run the TTS locally, but if you have a strong CPU, you can run a TTS also only in CPU mode, means only use normal system RAM, not VRAM. Some TTS are fast enough for CPU only, some are even too slow on GPU.