Microsoft Asia drops Mage Flow, a 4B DiT image model on Hugging Face
multimodalart · x · 2026-07-22
Microsoft Asia has released Mage Flow, a 4B-parameter DiT text-to-image and image-editing model on Hugging Face.
- The model can generate images from 512×512 up to 4K.
- The attached demo images suggest it can handle a range of styles, from landscapes and portraits to posters and stylized illustrations.
- The poster author says they built a demo and are impressed by the quality so far.
Related event: Microsoft Releases Open-Source 4B Image Model Mage-Flow(7 posts)→
More from Multimodal
- ChatGPT turns Amsterdam into a hamster coffee shop after a pun gone wrong — himanshustwts · 2026-07-22
- Midjourney 8.1 prompt shows a vivid snake-eye close-up with style reference settings — michaelrabone · 2026-07-22
- Creative Challenge #22 invites AI art made around a portal theme — azed_ai · 2026-07-22
- Cleanvoice launches Music Mixing 3.0 for speech-music separation and auto ducking — Scobleizer · 2026-07-22
- Seedance-generated city run demo lands on Taluna — ScheduleNo9532 · 2026-07-22
- Vivago Video Agent Beta Opens: AI Video Generation for E-commerce & Marketing — SimplyAnnisa · 2026-07-22