My favorite model list (OUTDATED)
Here I have collected models that I have enjoyed and that I actively use for RP or creativity
Updated • 147 • 9Note The best model I'm using right now. This is the first L3 model I have used in RP. Very good made and perfect for rp conversations. A great understanding of the context and an unforgettable experience in communication. With proper customization of instructions and character cards, this model can bring you to your knees. For me personally, it's a replacement for character.ai. My review can be found at the model link above. Creator @cgato Quant @Lewdiculous
Lewdiculous/Nyanade_Stunna-Maid-7B-v0.2-GGUF-IQ-Imatrix
Updated • 813 • 43Note Creative and interesting model, great for RP but requires proper settings. My favorite. Creator @ChaoticNeutrals Quant @Lewdiculous
SanjiWatsuki/Kunoichi-DPO-v2-7B
Text Generation • Updated • 723 • 80Note One of the best models from SanjiWatsuki. Pretty creative and smart, lots of purple prose in rp, char cards understanding is good, I use this model for creative and text. Creator @SanjiWatsuki
KatyTheCutie/LemonadeRP-4.5.3
Text Generation • Updated • 270 • 46Note A model that has become a breakthrough in rp, many are using this model for mergers because this model is very successful. Understanding of char cards is +-good. Very hot and creative, there is purple prose but not too much, it all depends on your text. Creator @KatyTheCutie
Lewdiculous/L3-8B-Stheno-v3.2-GGUF-IQ-Imatrix
Updated • 10.2k • 101Note Successful model compared to 3.1 version, there are problems with stopping text, solved by limiting the response token to 200 - 210 good balance. I also put a setting in SillyTavern with autocorrect markdown. On the plus side the model has become observant and respects context better, nice refreshing model. I suggest try it. Recommend using Virt-io presets and custom MinP at your discretion. My feedback can be found at the link above, in 'Community' section. Creator @Sao10K Quant @Lewdiculous
icefog72/WestIceLemonTeaRP-32k-7b
Text Generation • Updated • 40 • 10Note Just a good model, follows the character cards very well provided you customize them well. In the long run rp is a great option. Works in two formats Alpaca and ChatML, good text formatting.
Lewdiculous/LLaMa-3-CursedStock-v1.8-8B-GGUF-IQ-Imatrix-Request
Updated • 177 • 8Note I've never seen such a great merges in one model, surprisingly it's very good. Llama 3 formatting works great. (there are moments where the formatting can be broken but it is solved by swiping 90% of the time). With this model I seem to feel more like it's CAI style. Smart enough and creative enough and flexible. Creators @PJMixers Quant @Lewdiculous
Lewdiculous/L3.1-8B-Niitama-v1.1-GGUF-IQ-Imatrix
Updated • 782 • 16Note Someone said: "Better than Lunaris, which in turn was better than Stheno 3.2." I agree with this, when it comes to L3 Niitama v1. Specifically this model may be worse compared to the previous Niitama version, but it's pretty good too. Anyway, subjectively speaking, for me it's better than Stheno 3.2 in many ways. Try this, especially the Niitama v1 version and you probably won't regret it.