OccultAI/Qliphoth-12B-v1.1

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:12BQuant:FP8Ctx Length:32kPublished:May 23, 2026Architecture:Transformer0.0K Warm

OccultAI/Qliphoth-12B-v1.1 is a 12 billion parameter merged language model, combining multiple Mistral-Nemo-based models. Developed by OccultAI, it is trained on specialized datasets including Naphula-Archives/qliphoth_12B_minibard_bench and OccultAI/illuminati_imatrix_v1. This model is designed for diverse applications, leveraging the strengths of its constituent base models for broad utility.

Loading preview...

OccultAI/Qliphoth-12B-v1.1: Merged Language Model

Qliphoth-12B-v1.1 is a 12 billion parameter language model developed by OccultAI, created through a mergekit process. This model integrates the capabilities of numerous Mistral-Nemo-based models, aiming to combine their individual strengths into a single, versatile architecture. The merging approach allows it to draw from a wide array of pre-trained knowledge and fine-tuning objectives present in its base models.

Key Characteristics

  • Architecture: Based on the Mistral-Nemo family, indicating a robust foundation for general language understanding and generation tasks.
  • Parameter Count: Features 12 billion parameters, placing it in a capable size class for various applications.
  • Training Data: Utilizes specialized datasets such as Naphula-Archives/qliphoth_12B_minibard_bench and OccultAI/illuminati_imatrix_v1, suggesting a focus on specific domains or types of content.
  • Development Method: Constructed using mergekit, a technique for combining multiple models to enhance performance or broaden capabilities without extensive retraining.

Potential Use Cases

  • General Text Generation: Capable of generating coherent and contextually relevant text across a wide range of topics.
  • Exploratory AI Development: Suitable for developers looking to experiment with a model that synthesizes diverse characteristics from multiple base models.
  • Specialized Content Creation: Given its unique training datasets, it may excel in generating content related to the themes or domains covered by qliphoth_12B_minibard_bench and illuminati_imatrix_v1.