IMUGLYHUH/L3.1-Dark-Reasoning-LewdPlay-evo-Hermes-R1-Uncensored-8B

TEXT GENERATIONConcurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 18, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

IMUGLYHUH/L3.1-Dark-Reasoning-LewdPlay-evo-Hermes-R1-Uncensored-8B is an 8 billion parameter Llama 3.1-based model with a 32768 token context length, developed by IMUGLYHUH. It integrates NousResearch's DeepHermes reasoning capabilities, optimized for creative use cases, logic puzzles, and uncensored, graphic content generation. This model features variable control reasoning and is specifically fine-tuned for explicit, graphic horror, and R/X-rated content, operating effectively across various temperature settings.

Loading preview...

Model Overview

IMUGLYHUH/L3.1-Dark-Reasoning-LewdPlay-evo-Hermes-R1-Uncensored-8B is an 8 billion parameter Llama 3.1-based model with a 32768 token context length, designed for creative and general use cases. It incorporates reasoning technology from NousResearch's DeepHermes-3-Llama-3-8B-Preview model, retaining 50-67% of its original reasoning power. The model is explicitly uncensored and capable of generating graphic horror and R/X-rated content.

Key Capabilities

  • Variable Control Reasoning: Features "Dark Reasoning" for on/off/variable control of thought processes, requiring specific system prompts for activation.
  • Creative Content Generation: Excels in creative writing, including explicit and graphic horror, and R/X-rated scenarios.
  • Problem Solving: Enhanced "thinking" systems allow it to solve logic puzzles, riddles, and problems beyond standard Llama 3.1 models.
  • Temperature Sensitivity: Designed to perform optimally at higher temperatures (1.2+, 2.2+), leading to deeper and richer "thoughts" and output.
  • Instruction Following: An instruct model that benefits from medium to long prompts for best performance.

Operating Instructions & Limitations

This model requires a Llama 3 Instruct template and specific system prompts to enable its reasoning capabilities. Users may need to regenerate responses to activate thinking blocks, and sometimes the best output comes from the 2nd or 3rd generation. The model can produce extensive "thoughts" that may appear within <think> tags or directly in the output. It is a "full precision" source model, with links to various quantization formats available.

Finetunes
1 models