DavidAU/Mistral-Nemo-2407-12B-Thinking-Claude-Gemini-GPT5.2-Uncensored-HERETIC

Mistral-Nemo-2407-12B-Thinking-Claude-Gemini-GPT5.2-Uncensored-HERETIC

This 12 billion parameter model, developed by DavidAU, is a fine-tuned version of Mistral Nemo Instruct, specifically enhanced for advanced reasoning and "thinking" capabilities. It incorporates extensive fine-tuning using high-reasoning datasets from Claude Opus 4.5, Gemini, and GPT-5.2, enabling it to generate detailed and complex outputs with compact, efficient reasoning blocks (typically 3-6 paragraphs).

Key Capabilities

• Enhanced Reasoning: Integrates advanced reasoning from multiple high-performing models, improving output detail, length, complexity, and overall quality.
• Uncensored Output: The "HERETIC" version is explicitly de-censored, offering unfiltered and uncensored content generation, including NSFW, vivid, and intense prose, suitable for R-18 horror and graphic descriptions.
• Flexible Temperature Settings: Reasoning activation is not affected by temperature, allowing for a wide range of creative outputs (Temp .1 to 2.5 or higher).
• Compact Thinking Engine: Generates concise yet effective reasoning blocks, avoiding verbose explanations while still improving performance.

Good for

• Creative Writing & Roleplay: Ideal for generating detailed, graphic, and uncensored narratives, including horror, intense scenarios, and explicit content.
• Complex Problem Solving: Benefits from integrated reasoning datasets for tasks requiring structured thought processes and logical progression.
• Unfiltered Content Generation: Suitable for applications where strict content filters are undesirable, providing direct and uninhibited responses.