Sorihon/CYDR-24B

TEXT GENERATIONConcurrency Cost:2Model Size:24BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Jun 9, 2026Architecture:Transformer Cold

Sorihon/CYDR-24B is a 24 billion parameter merged language model developed by Sorihon, built from a combination of several 24B models including Sorihon/Magistry-Painted-Cydonia-24B and Delta-Vector/Rei-24B-KTO. This model is specifically designed for adaptable roleplay scenarios, excelling at prompt following and providing short, concise replies. It features a 32768 token context length and is optimized for interactive narrative generation.

Loading preview...

Sorihon/CYDR-24B: A Roleplay-Optimized Merged Model

Sorihon/CYDR-24B is a 24 billion parameter language model created by merging several specialized 24B models, including Sorihon/Magistry-Painted-Cydonia-24B, Delta-Vector/Rei-24B-KTO, TheDrummer/Cydonia-24B-v4.3, Entropicengine/Pinecone-sage-24b, and MuXodious/WeirdCompound-v1.7-24b-absolute-heresy. Its primary design goal is to provide a model highly adaptable for various roleplay scenarios, with a preference for short, direct responses unless prompted otherwise.

Key Characteristics & Capabilities

  • Roleplay Optimization: Specifically engineered for dynamic and adaptable roleplaying, demonstrating strong prompt following.
  • Concise Responses: Out-of-the-box, the model prefers short replies, making it suitable for quick, interactive exchanges.
  • Prompt Following: Exhibits good prompt adherence, though this can sometimes lead to overly literal interpretations.
  • Context Length: Supports a substantial context window of 32768 tokens, allowing for more detailed scenario management.

Potential Use Cases

  • Interactive Roleplay: Ideal for engaging in quick, scenario-based roleplaying with decent results.
  • Character Dialogue Generation: Can be used to generate short, character-specific responses in narrative contexts.
  • Prompt-Driven Content: Suitable for applications where strict adherence to prompt instructions is crucial, even if it requires more detailed prompting.

Known Limitations

  • Repetition Loops: May occasionally fall into repetition, especially with minimal input or specific system prompt structures.
  • Literal Interpretation: Can interpret prompts very literally, potentially leading to unexpected or frustrating outcomes.
  • Language Support: Primarily tested and confirmed for English language use.

Finetunes
2 models