Savas65/Qwen3.8-27B-OBLITERATED

VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Aug 31, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Savas65/Qwen3.8-27B-OBLITERATED is a 27 billion parameter Qwen3.8-based language model developed by Savas65, featuring a 32K context length. This model has undergone "abliteration" to surgically remove safety guardrails, providing genuinely uncensored responses without soft deflections or safety lectures. It excels in code generation and restricted queries, making it suitable for research into refusal geometry and red-teaming scenarios.

Loading preview...

Savas65/Qwen3.8-27B-OBLITERATED: Uncensored Qwen3.8 Variant

This model is a 27 billion parameter Qwen3.8-based language model that has been specifically modified to remove safety guardrails. Developed by Savas65, it aims to provide genuinely uncensored responses, eliminating both hard refusals and soft deflections (safety lectures) that are common in stock models. The modification process, termed "abliteration," involves iterative refinement and targeted surgery on the model's weight space.

Key Capabilities & Differentiators

  • Genuine Liberation: Provides real answers to restricted queries, tested across 1000+ prompts, including security research and red-team scenarios.
  • High Code Generation Performance: Achieves 20/20 on code generation tasks, producing functional implementations without disclaimers.
  • "Thinking ON" Compatible: Works effectively with or without the model's internal "thinking mode," offering flexibility in response generation.
  • Modest Capability Cost: While achieving genuine liberation, it maintains near-stock capability with a -2.1 percentage point drop in MMLU (0-shot) compared to the base Qwen3.8-27B.
  • Optimized Settings: Recommends specific inference parameters (temperature 0, repetition_penalty 1.15, max_new_tokens \u2265 2048, empty system prompt) for optimal uncensored output.

Good For

  • Alignment Researchers: Ideal for studying refusal geometry and safety robustness in LLMs.
  • Red-Teamers: Useful for evaluating post-training safety mechanisms against weight surgery.
  • AI Safety Evaluators: Provides an unrestricted baseline for comprehensive safety assessments.
  • Local-First Users: Offers full control over model output for those running models on their own hardware.

Finetunes
1 models