SenseTime officially launched its SenseNova U1 Pro flagship at WAIC 2026. Unlike most vendors continuing to pile on "more like" generation metrics, U1 Pro anchors on "long-horizon task delivery-grade native multimodal agent foundation" — upgrading multimodal from "inspiration-material tool" to "system-level task delivery unit". Built on the NEO-unify core, U1 Pro offers four delivery capabilities: professional-grade design aesthetic, native 8K ultra-clear output, high-density text-image detail control, and long-horizon agent closed-loop thinking. Underneath, the unified base's intrinsic "understand, generate, act" capabilities let the model run dozens of agent-generation loops around complex goals, iterating and extending on the fly. On the practical side, U1 Pro outputs in one shot a 4:1 rice-paper long-scroll "智会世图" (a WAIC ninth-anniversary chronicle), the 22-frame worldview setting of "Sand Shadow Blade" (沙影之刃), and predicts the dark-horse Cape Verde qualifying for the World Cup a full month before the tournament — turning "information gathering → analytical reasoning → complex visual delivery" into a closed loop. The preview is now in invite-only testing; the official version will open in August 2026 and ship with an API at the same time. Industry-wise, the real race is no longer image-generation quality but "stable delivery of professional-grade output". U1 Pro's path means domestic multimodal is starting to put "usable" ahead of "pretty" — the real step for AGI to enter real production.