MS3.2-24b-Angel
Base: Mistral-Small-3.2-24B-Instruct-2506 • by Fizz
A model finetuned from Mistral Small 3.2 for roleplaying, storywriting, and general instruct usecases. Strong prose and character portrayal rivaling 72B models.
Links:
Full |
EXL3 |
GGUF
|
Llama-3.3-70B Joyous
Base: meta-llama/Llama-3.3-70B-Instruct • by Fizz
A finetune of Llama 3.3 70B designed for roleplay that also turned out to be comically good at assistant tasks in subjective evals.
Links:
Full
|
Luna-27B v0
Base: YanLabs/gemma-3-27b-it-abliterated-normpreserve-v1 • by Fizz
A Gemma 3 27B finetune (trained on top of an abliterated base) originally intended for roleplay, with unexpectedly excellent assistant capability (especially media analysis and self-awareness tasks).
Links:
Full
|
Tlacuilo-12B
Base: LatitudeGames/Muse-12B (Mistral-Nemo-Base-2407 family) • by Toasty Pigeon
Creative writing model with improved RP/adventure performance vs recent iterations, tuned in staged training (books → RP → small instruct phase).
Links:
Full
|
|