The 2026 Artificial Intelligence Industry Conference opened from October 14 to 16 at the Jinan Yellow River International Convention and Exhibition Center, with the opening ceremony and main forum on October 15. Jointly hosted by the China Academy of Information and Communications Technology (CAICT), the Shandong Provincial Department of Industry and Information Technology and the Jinan Municipal Government, the conference is themed "Accelerating AI Empowering New Industrialization."
At the conference, seven vertical large models debuted together — the "Qilu Seven Stars" — industrial-grade AI products born from real workshops and rooted in production lines. Unlike general-purpose models built for chat and content, these focus on core industrial pain points: whether a line runs more smoothly, costs fall further and risks are caught earlier.
Inspur's Hairuo model is Shandong's first vertical product to pass the Cyberspace Administration's generative-AI service filing. Staff said that, relying on a standardized model production line, it compresses the industry-model development cycle from 90 days to 20 days. In petrochemical scenarios it extends a plant's overhaul cycle from two to three years to three and a half years, saving enterprises tens of millions of yuan annually.
COSMO's Tianszhi model, rooted in home appliances and process chemicals, integrates production data scattered across a dozen independent systems and compresses flexible line scheduling from hours to minute-level automatic output. The Jixianguanjiao interstellar vision-language model, built on 1 billion real labeled data from 6,000-plus industrial projects, needs no per-scene small-model training and reaches over 90% accuracy in unfamiliar scenarios across 100-plus industries. CloudRock's Shanhai model, focused on smart mines, has found more than 10,000 potential safety hazards.
Source: Qilu Evening News / Jinan Government (English)
