Qwen-Image-2.1 uncensored GGUF hits Hugging Face for local ComfyUI image generation
prajdabre · x · 2026-09-21
Hugging Face user abenzerps has released uncensored GGUF quantizations of Qwen-Image-2.1, enabling local text-to-image generation via ComfyUI + ComfyUI-GGUF.
- Quantizations from Q40 to Q80; Q4KM (4.6GB) is recommended as the best size/quality balance, while Q80 may hit a shape mismatch error on some GPUs
- Companion text encoders (qwen3vl8b in BF16 or Int8) and VAE files are packaged in the repo with full setup instructions
- The poster quips that a model trained for millions of dollars to SoTA standards gets uncensored by the community almost immediately after release
Related event: Qwen-Image-2.1 GGUF Quantized Version Tops Hugging Face Trending(4 posts)→
More from Multimodal
- Redditor benchmarks Minimax H3 Image samplers side-by-side at 16MP+ with identical jaguar prompt — Phantom-log-9484 · 2026-11-12
- Indie Britpop AI music video built with Nano Banana Pro, Kling, and free lip-sync via WAN2GP + LTX 2.5 — cocktailpeanut · 2026-11-12
- P2Echo: one text-prompted model segments multi-view echocardiography with high accuracy and low cost — UBC-V · 2026-11-12
- Redditor makes fast-motion short film with H3 and ships Longshot Studio v0.18.5 with face swap — R34vspec · 2026-11-12
- Reimagining Knight Rider's opening as a Netflix drama with Claude Fable 5.2 — justin_hart · 2026-11-12
- Gemini 4 Carbon builds a full 3D aquarium in Blender and Three.js in ~6 hours — sunjiao123sun_ · 2026-11-12