Abliterated Qwen-Image-2.1 drops refusals from 100/100 to 5, runs locally at 4.68GB
alexcovo_eth · x · 2026-09-23
A community release of an abliterated Qwen-Image-2.1 claims a real fix this time: the uncensoring is done in the text encoder rather than the image weights, so normal prompts don't degrade. Refusals drop from 100/100 to 5/100. The Q4KM GGUF is only 4.68GB and runs on a laptop via ComfyUI with city96's GGUF loader. It also supports editing with up to 10 reference images and native transparency output for stickers/cutouts without a masking step. Model is on Hugging Face.
More from Multimodal
- Third-party test: Claude Opus 5.5 renders finer 3D scenes but costs 13x more than GPT-6 Sol — testingcatalog · 2026-09-23
- ComfyUI trick: aux preprocessor + Qwen transfers poses across characters with one prompt — Acceptable-Work8202 · 2026-09-23
- Same portrait prompt across Midjourney V6.1, V7 and V8.2: do older models look better? — tisch_eins · 2026-09-23
- Testing AI character consistency across a 20-image travel sequence — SiennaVaire · 2026-09-23
- Midjourney v8.2 Faces: New Portrait Generation Samples Shared — azed_ai · 2026-09-23
- One-sentence prompt generates lifelike dog video, shown side-by-side with the real one — wgrathwohl · 2026-09-23