TheDrummer/Cydonia-R1-24B-v4

Hugging Face
TEXT GENERATIONConcurrency Cost:2Model Size:24BQuant:FP8Ctx Length:32kPublished:Aug 1, 2025Architecture:Transformer0.0K Warm

TheDrummer/Cydonia-R1-24B-v4 is a 24 billion parameter language model based on the Mistral v7 Tekken architecture, developed by TheDrummer. This model is noted for its creative writing capabilities, ability to maintain narrative consistency with multiple characters, and strong reasoning, making it suitable for complex storytelling and roleplay applications. It demonstrates solid response quality and engaging internal thought processes.

Loading preview...

Cydonia R1 24B v4 Overview

Cydonia R1 24B v4, developed by TheDrummer, is a 24 billion parameter language model built upon the Mistral v7 Tekken architecture. This iteration is highlighted for its robust performance in creative writing and complex narrative generation, with user feedback emphasizing its ability to maintain story continuity and character consistency across extended interactions. The model is also noted for its strong reasoning capabilities, often exhibiting engaging internal thought processes that contribute to its nuanced responses.

Key Capabilities

  • Creative Writing: Excels at generating detailed and imaginative content, fleshing out replies comprehensively.
  • Narrative Consistency: Demonstrates proficiency in continuing ongoing stories and tracking multiple characters without repetition.
  • Reasoning: Provides insightful reasoning, with users noting its "smart" and "interesting" internal thoughts.
  • Character Emulation: Capable of capturing and maintaining character personas effectively.

Usage Notes

  • Users may need to prefill prompts (e.g., <think>\nOkay,) to optimize its reasoning output.
  • It is compatible with various formats, including GGUF, iMatrix (recommended), and EXL3.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p