>>542628910
Yeah, I saw people in here hate "chatshit" so I was unsure. I used to write a lot with NAI last year or so, but stopped because the 8k context was just too low. I've heard they have a new model with larger context now though, so might try it out again.

>>542628961
>Is deepseek worth trying to set it up?
It's what I use almost exclusively. DS is super cheap and (almost) completely uncensored. The newest model (V3.2 exp) seems like a good improvement over the previous ones too. If you want to try it yourself it's important you use the -reasoner version though, not the -chat version. Because the devs fucked up and somehow killed the context comprehension on exp-chat.

I don't know if you guys here know about the Fiction.Live benchmark. They test how good a LLM can actually comprehend large context sizes, which is important for RP and creative writing. Pic related are the DS results. Here's the full table with all tested models: https://fiction.live/stories/Fiction-liveBench-Sept-29-2025/oQdzQvKHw8JyXbN87/home

>>106819110
Mega updated.
https://mega.nz/folder/KGxn3DYS#ZpvxbkJ8AxF7mxqLqTQV1w
Rentry updated with new main prompt example and suggestion not to use -chat based on poor context memory.