OccultAI/Qliphoth-12B-v1

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:12BQuant:FP8Ctx Length:32kPublished:May 22, 2026Architecture:Transformer0.0K Warm

OccultAI/Qliphoth-12B-v1 is a 12 billion parameter merged language model, combining multiple Mistral-Nemo-based models. Developed by OccultAI, this model integrates diverse capabilities from its constituent models, including reasoning, roleplay, and instruction following. It is designed for general-purpose applications where a blend of these strengths is beneficial. The model leverages a mergekit architecture to synthesize varied fine-tuning objectives.

Loading preview...

OccultAI/Qliphoth-12B-v1 Overview

OccultAI/Qliphoth-12B-v1 is a 12 billion parameter language model created by OccultAI, built using a mergekit architecture. This model is a sophisticated blend of numerous base models, primarily derived from the Mistral-Nemo-Instruct-2407 family and various fine-tuned derivatives. The merging process aims to synthesize a wide array of capabilities present in its constituent models.

Key Characteristics

  • Merged Architecture: Combines over 30 distinct 12B parameter models, including those focused on reasoning, roleplay, instruction following, and specialized domains like psychology and religious texts.
  • Diverse Training Influences: Integrates fine-tuning from models such as allura-org/Tlacuilo-12B, anthracite-org/magnum-v4-12b, ChaoticNeutrals/Mag-Mell-Reasoner-12B, PygmalionAI/Pygmalion-3-12B, and sleepdeprived3/Christian-Bible-Expert-v2.0-12B, among many others.
  • General-Purpose Utility: Designed to offer a broad spectrum of capabilities by drawing from the strengths of its diverse merged components.

Potential Use Cases

  • General Instruction Following: Capable of handling a variety of prompts due to its broad training base.
  • Creative Writing and Roleplay: Benefits from the inclusion of models specifically fine-tuned for these tasks.
  • Reasoning Tasks: Incorporates models known for their reasoning capabilities, suggesting potential for complex problem-solving.
  • Exploration of Niche Domains: The diverse merge includes models with specialized knowledge, which might be leveraged for specific applications.

Finetunes
2 models