At the World Artificial Intelligence Conference in Shanghai (17–20 July 2026), Suzhou-based UniX AI put its wheeled dual-arm Panther humanoid on a kitchen line that Chinese trade coverage describes as a continuous pizza workflow — from working the dough through sauce, toppings, oven time, slicing, boxing and handoff — rather than a single pick-and-place gag.
Reporting on the booth frames the run as long-horizon task organization: UniCortex decomposes “make a pizza” into staged skills and can call learned motions plus oven-side IoT steps; UniFlex supplies imitation-learned skills such as spreading sauce and dual-arm tool use; UniTouch adds visuo-tactile contact cues. Xinhua Daily Finance’s show-floor colour says visitors waited for pies that went from raw to cooked in about eight minutes, and an incomplete tally put roughly 140 pizzas (plus sugar-painting demos) through the stand by the morning of 20 July.
This story links to UniX AI and the Panther — a full-size wheeled dual-arm service platform (OEM page: omnidirectional chassis, mass-produced 8-DoF arms, UniCortex/UniFlex/UniTouch stack). Treat the piece as a deployment-demo update for that line, not as a new SKU launch.
WAIC is a staged exhibition kitchen with show traffic, not an independent restaurant throughput study. Volume and cycle-time figures are booth journalism, not audited SLA. UniX marketing elsewhere uses stronger “first household / mass-produced” language for Panther; keep those claims out of this pizza story unless you are writing the April home-deployment announcement. A mobile-app “order then cook” thread circulating in informal notes is not pinned in the English/Chinese trade pieces cited here, so it is omitted as fact.