
These models are quite good
Text Generation • 24B • Updated • 47 • 17Note ⭐Follow char's persona well.
trashpanda-org/Llama3-24B-Mullein-v1
24B • Updated • 66 • 8Note Quite creative, diverse language use (not very AI-ish), but tends to break format/speaking user/leaking sysprompt or author note into chat
redrix/GodSlayer-12B-ABYSS
Text Generation • 12B • Updated • 53 • 16Note Quite creative, casual language use (not very AI-ish), but tends to break format/speaking user/leaking sysprompt or author note into chat
lars1234/Mistral-Small-24B-Instruct-2501-writer
24B • Updated • 46 • 19Note Very strict on following char's persona. Max tokens settings don't seem to work and it prefers to gives long paragraphs of output. Somewhat repetitive even when increasing temperature, but XTC helps a bit.
yamatazen/EtherealAurora-12B-v2
Text Generation • 12B • Updated • 175 • • 27Note Prose choice still noticeably has that purple style, but this one is better (no leaking sysprompt/AN). Follows max token settings well. *Tried seducing a char by doing physical stuff, and they maintained their persona (dialogues did not fall quickly into moans, groans, etc.)
Gryphe/Pantheon-RP-1.8-24b-Small-3.1
24B • Updated • 125 • 66Note Follows char's persona well, max token settings work, prose still has a little bit of AI slop. Just below apparatus 24b for me.