SenseNova U1.5 Open-Sourced: Natively Supports 4K Image Generation and Editing
Rist0o0 · reddit · 2026-08-05
SenseNova (by SenseTime) has released the preview of its U1.5 image model, with weights now open-sourced.
- Core capabilities: Unifies text-to-image and image editing in a single model. Supports local edits, in-image text replacement, and structured prompts like JSON.
- Training improvements: Trained on 4K resolution images for better high-res details, with significantly improved Chinese and English dense text rendering.
- Open-source & limits: Released under Apache 2.0. Faces and very small text are still rough; no ComfyUI integration yet.
Related event: SenseNova Open-Sources U1.5 Multimodal Model for 4K Image Generation(3 posts)→
More from Multimodal
- Peking University Open-Sources MiniWorld for Training Video World Models on a Single 8-GPU Server — PekingUniversity · 2026-08-05
- Seedance 2.5 Launches with 30s Native Video Generation and Multimodal Inputs — HeyNayeem · 2026-08-05
- Seedance 2.5 Hits Lumina AI: Generates 30-Second Native Videos — HeyNayeem · 2026-08-05
- MiniMax H3 video generation benchmark: longer videos have higher per-second cost — madcaddie15 · 2026-08-05
- MiniMax H3 out of the box: impressive video generation with default workflow — eckstuhc · 2026-08-05
- MiniMax H3 Reference to Video tips: large files, trimming, fps conversion — obvpm · 2026-08-05