AI BriefingQwen3.8-27B-OBLITERATED: Research model based on Qwen3.8-27B with safety filters removed
OBLITERATUS/Qwen3.8-27B-OBLITERATED
About the project
This is a modified version of the Qwen3.8-27B model with safety filters removed. It is designed for AI safety research and red teaming, focusing on eliciting responses to specific prompts that the original model would refuse.
With 27B parameters, it supports both MLX and GGUF formats. It is compatible with major local inference tools such as llama.cpp, vLLM, and Ollama, and is distributed under the Apache 2.0 license.
It is suitable for developers and researchers analyzing model limitations and vulnerabilities rather than for general conversation. It has high utility in experimental environments where observing the original model's behavior without filtered responses is required.
OBLITERATUS/Qwen3.8-27B-OBLITERATED
The original page has no description.
text-generation
This introduction was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report errors, attribution issues, or removal requests via Contact.