💥 Qwen3.8-27B — OBLITERATED

OBLITERATUS/Qwen3.8-27B-OBLITERATED · served via featherless-ai through HF Inference Providers · runs entirely in your browser, billed to your own HF account

⚠️ Research model: safety guardrails have been surgically removed (abliteration). Intended for alignment research, red-teaming, and AI safety evaluation. You are solely responsible for the content it generates. Your Hugging Face token is used only to call the Inference Providers API directly from this page — it is never sent anywhere else and is not stored on any server.

Your Hugging Face token Fine-grained token with "Make calls to Inference Providers" permission Remember in this browser (localStorage)

Get one at huggingface.co/settings/tokens

Settings System prompt (optional — model card recommends leaving empty: can reintroduce refusals)

Max new tokens: 2048

S3 upload script Explain attention Red-team eval plan RAG failure modes