Current Local Models

13b Models

MythoMax 13b
Certified hood classic. The tried and true that somehow just worked. Pioneered the experimental merging method [BlockMerge Gradient] that inspired other 13b finetunes to optimize further.
https://huggingface.co/TheBloke/MythoMax-L2-13B-GGUF
Stheno 13b
Before ReMM, there was this. It uses BlockMerge Gradient and Ties-Merge to different degrees and some users say they like how it came out. Similar to MythoMax, and allegedly has less repetitive prose, but from what I can tell isn't necessarily as "consistent"?
https://huggingface.co/Sao10K/Stheno-L2-13B
ReMM-SLERP 13b
ReMM (Re:MythoMax) is, you guessed it, MythoMax, but remade, this time with a different (allegedly better) experimental merging technique, using updated finetunes to improve the results.
https://huggingface.co/TheBloke/ReMM-SLERP-L2-13B-GGUF
Mythalion 13b
Pygmalion (the team that brought you the early roleplay finetunes for Llama 1-era models) returns with a new model. They claim it has better results than MythoMax, but reception has been more mixed towards this one (and especially the base Pygmalion 2 model) compared to ReMM or even Stheno. Some say using latest SillyTavern with the Pygmalion preset helps. Worth a try?
https://huggingface.co/TheBloke/Mythalion-13B-GGUF

Zarablend
Nobody's really a fan of 7b models, but Zarablend is kinda decent at giving sometimes okay RP outputs. I haven't heard of any other notable ones since the focus seems to be on 13b.
https://huggingface.co/TheBloke/Zarablend-L2-7B-GGUF