MOSS-Image-Edit-HQ v5 lands on Hugging Face for instruction-based image editing
Illustrious_Row_9971 · reddit · 2026-10-25
MOSS-Image-Edit-HQ v5 has been released on Hugging Face (akhaliq/MOSS-Image-Edit-HQ), positioned as a high-quality instruction-based image editing model.
Samples show natural-language edits like changing a background to a snowy winter evening while preserving the subject's details.
More from Multimodal
- Moss Image demos turn heads on Reddit: a 5B single-stream DiT trained with Muon — lovelycontributer · 2026-10-25
- Ant-affiliated inclusionAI's 0.5B open-source TTS model trends on Hugging Face — inclusionAI · 2026-10-25
- One-take AI video: jet ski rescue through a drowning city with no cuts — umesh_ai · 2026-10-25
- Ribbed-glass + neon lighting: a stylized Clayface Midjourney prompt — michaelrabone · 2026-10-25
- Midjourney explorations in a 1930s black-and-white cinematic style — hewarsaber · 2026-10-25
- FIBO 2 vs FLUX 3 Klein side-by-side: photorealistic mug shelf render-off — RobbaW · 2026-10-25