huoaaa/Stellar-Witch-12B

TEXT GENERATIONPricing:Input $1.2 / Output $4.8Concurrent Unit Cost:1Model Size:12.2BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 5, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Stellar-Witch-12B is a 12.2 billion parameter language model created by huoaaa through a custom merge of five distinct 12B models: Stellar-Seraph-12B, MN-12B-Mag-Mell-R1, Dans-SakuraKaze-V1.0.0-12b, NeonMaid-12B-v2, and Ollpheist-12B. This model is specifically optimized for creative writing tasks, excelling in roleplay, storytelling, and generating atmospheric long-form narratives. It features a 32768 token context length, making it suitable for extended creative applications.

Loading preview...

Stellar-Witch-12B: Merged Model for Creative Writing

Stellar-Witch-12B is a 12.2 billion parameter language model developed by huoaaa. It was created using a custom merge method, combining the strengths of five different 12B models: Stellar-Seraph-12B, MN-12B-Mag-Mell-R1, Dans-SakuraKaze-V1.0.0-12b, NeonMaid-12B-v2, and Ollpheist-12B. The merge utilized specific parameters for strength, prose, gravity, and adherence, with Vortex5/Stellar-Seraph-12B serving as the base model and tokenizer source.

Key Capabilities

  • Roleplay: Designed for emotion-forward interactive scenarios.
  • Storytelling: Excels at generating atmospheric and long-form narratives.
  • Creative Writing: Optimized for producing atmospheric fiction.

Intended Use

This model is specifically tailored for applications requiring high-quality, imaginative text generation, particularly in the domains of interactive role-playing, crafting detailed stories, and general creative writing. Its merged architecture aims to enhance its ability to produce engaging and atmospheric content.

Finetunes
1 models