SenseNova U1Pro Review: Multi-Round Edits Without Drift Push AI Image Generation Toward Deliverable Output

量子位 · wechat · 2026-09-22

SenseNova U1Pro has officially launched on the Xiaohuanxiong platform with enterprise API access. QbitAI's hands-on tests show near-photorealistic output and, more importantly, deliverability: precise local edits, multi-reference-image composition (four images merged into a cat-cafe miniature poster), and three consecutive edit rounds — changing price, headline and cup color — without disturbing the rest of the image.

It can also generate 8-panel continuous action sequences convertible to GIF for video pre-production, autonomously research NASA sources to build accurate Chinese infographic posters, and produce e-commerce scene shots. SenseNow integrates requirement understanding, retrieval, asset handling, generation and correction into one creative pipeline.

Related event: SenseTime's SenseNova U1Pro Tested: Image Generation Reaches Deliverable Quality(2 posts)→

Original post →

More from Multimodal

Multimodal channel →