Ateron/Gemma-4-Dark-Thoughts-31B

Hugging Face
VISIONConcurrent Unit Cost:2Model Size:31BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 7, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Warm

Ateron/Gemma-4-Dark-Thoughts-31B is a 31 billion parameter language model based on the Gemma-4 architecture, created by Ateron through a series of model merges. This model is specifically optimized for role-play and creative writing tasks, demonstrating an ability to easily adopt system prompts and maintain story direction without passiveness. It leverages a unique two-phase mergekit configuration, combining models like MeroMero V2, Dark-Scarlett V2, and Scotoma v2 to enhance its creative generation capabilities. With a 32768 token context length, it is well-suited for extended narrative generation and complex character interactions.

Loading preview...

Overview

Ateron/Gemma-4-Dark-Thoughts-31B is a 31 billion parameter merged language model, specifically designed and optimized for role-play and creative writing. Developed by Ateron, this model is noted for its ability to readily interpret and follow system prompts, maintaining narrative direction without exhibiting passiveness, which is crucial for engaging and consistent storytelling.

Key Capabilities

  • Enhanced Role-Play: Excels at adopting personas and following complex character interactions within a narrative.
  • Creative Writing: Optimized for generating imaginative and coherent stories, dialogues, and descriptive text.
  • Prompt Adherence: Demonstrates strong capability in understanding and adhering to system prompts and story directions.
  • Mergekit Architecture: Built using a two-phase dare_ties mergekit configuration, combining several high-performing base models including Gemma-4-MeroMero-V2, Gemma-4-Dark-Scarlett-V2, and Gemma-4-Scotoma-V2 to achieve its specialized performance.

Good For

  • Developers and writers looking for a model capable of generating high-quality, consistent content for role-playing scenarios.
  • Applications requiring creative text generation where adherence to specific narrative guidelines and character traits is important.
  • Experimentation with merged models for specialized generative tasks, particularly within the creative domain.

Finetunes
2 models