SSDD145/Mistral-Nemo-2407-12B-Thinking-Claude-Gemini-GPT5.2-Uncensored-HERETIC

Hugging Face
TEXT GENERATIONPricing:Input $0.87 / Cached $0.2 / Output $0.99Concurrent Unit Cost:1Model Size:12BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Oct 1, 2026Architecture:Transformer Featherless Exclusive Warm

SSDD145/Mistral-Nemo-2407-12B-Thinking-Claude-Gemini-GPT5.2-Uncensored-HERETIC is a 12 billion parameter Mistral Nemo Instruct model, fine-tuned with Claude Opus 4.5, Gemini, and GPT5.2 datasets for enhanced reasoning and 'thinking' capabilities. It supports a context length of up to 1 million tokens, though 128k-256k is typical. This model is explicitly uncensored and unfiltered, designed to provide direct, vivid, and intense outputs, including R-18 horror and strong language, making it suitable for creative writing and roleplay without refusal.

Loading preview...

SSDD145/Mistral-Nemo-2407-12B-Thinking-Claude-Gemini-GPT5.2-Uncensored-HERETIC

This model is a 12 billion parameter variant of the Mistral Nemo Instruct architecture, significantly enhanced for advanced reasoning and 'thinking' processes. It has been fine-tuned using high-reasoning datasets from Claude Opus 4.5, Gemini, and GPT5.2, enabling it to generate detailed, multi-paragraph reasoning blocks (300-600 tokens) that improve overall output quality, detail, length, and complexity. The model supports an extended context window, typically 128k-256k tokens, with a maximum capacity of 1 million tokens.

Key Differentiators & Capabilities

  • Uncensored & Unfiltered: This "HERETIC" version is explicitly designed to be uncensored, providing direct and unfiltered responses, including vivid, intense, and R-18 content, swearing, and horror, without refusal. Its refusal rate is significantly reduced (14/100 compared to 87/100).
  • Enhanced Reasoning: Integrates a compact reasoning/thinking engine that self-generates internal thought processes, directly improving the quality and depth of generated text.
  • Context Length: Supports a substantial context window, crucial for complex tasks and maintaining coherence over long interactions.
  • Robust Output: Reasoning activation is not affected by temperature settings, allowing for flexible generation styles (Temp .1 to 2.5 or higher) while maintaining reasoning quality.

Use Cases

  • Creative Writing & Roleplay: Ideal for generating vivid, detailed, and uncensored narratives, including horror, romance, and other genres requiring intense or explicit content.
  • Complex Problem Solving: Benefits from its enhanced reasoning capabilities for tasks requiring multi-step thought processes and detailed explanations.
  • Applications Requiring Directness: Suitable for scenarios where unfiltered and straightforward responses are preferred, without built-in content moderation.

Note: Benchmarks provided are for the original Mistral Nemo model and do not yet reflect the impact of the reasoning fine-tuning.

Finetunes
1 models