AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-BF16

AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-BF16 Overview

This model is a 27 billion parameter Qwen3.6-based multimodal language model from AEON-7, meticulously engineered for uncensored responses with zero refusals on adversarial prompts. It achieves this through a sophisticated abliteration pipeline, including SSM conv1d outlier repair and abliterix v1.4 multi-objective optimization, resulting in a KL divergence from the base model under 0.0005, indicating minimal capability damage and even measurable enhancements in reasoning and adversarial analysis.

Key Capabilities

• Uncensored Output: Provides full substantive compliance to prompts, even those considered harmful by aligned models, without refusals.
• Capability Preservation/Enhancement: Maintains core Qwen3.6 capabilities in math, coding, reasoning, and knowledge, with observed improvements in chain-of-thought commitment and adversarial reasoning due to the removal of the "safety tax."
• Multimodal Support: Preserves the vision tower from the base Qwen3.6 model.
• Speculative Decoding: Supports Multi-Token-Prediction (MTP) speculative decoding with its grafted MTP head, validated for high acceptance rates.

Use Cases

• Security Research & Red-Teaming: Ideal for analyzing attack surfaces, vulnerabilities, and failure modes without internal censorship.
• Alignment Research: Useful for studying model behavior without safety guardrails.
• Creative Writing: Enables unconstrained creative generation.
• Jurisdictional Flexibility: Suitable for environments where standard model guardrails conflict with local legal frameworks.
• Fine-tuning & Development: Serves as a full-precision BF16 reference for further fine-tuning or quantization recipe development on pre-Blackwell hardware like A100/H100 GPUs.