The 2nd Comprehensive Transportation Large-Model AI Agent Contest closed in Mianyang, Sichuan on October 20. Team registrations, participants and valid entries all doubled year-over-year — "100+ scenarios, 1,000+ works, 10,000+ participants" — as China's transport sector pushes LLM agents from concept to frontline practice.
300 finalist works spanning all transport domains
Organized by the Ministry of Transport's Occupational Qualification Center with China Transport Information & Technology among the co-organizers, around 300 works reached the two-day finals. Ten entries — including a multi-scenario urban-rural logistics operations agent, a highway-network "digital duty officer", and a "BiaoZhiTong" standards-drafting agent — won Special Recommendation Awards, alongside 35 first prizes, 53 second prizes, 82 third prizes and 111 excellence awards.
Key numbers
· Registrations, participants and valid entries all doubled vs. the first edition
· 10 special recommendation + 35 first + 53 second + 82 third + 111 excellence awards
· First-ever "AI-evaluates-AI" testing and an international track
At the closing ceremony, Vice Minister Li Yang stressed three priorities: scenario-driven application so agents "take root on the front line"; collaborative development and shared results — "one breakthrough, many beneficiaries" — to cut duplicated R&D costs; and continued innovation and openness. At the opening, MOT Science & Technology Department head Xu Wenqiang pledged "prudent inclusiveness, risk-orientation, agile governance and tech-management integration" under the "AI + Transportation" application safety guidelines.
The bigger story: Comprehensive Transport Large Model 1.0
The ceremony also released the Comprehensive Transportation Large Model 1.0, the 2026 "AI + Transportation" Innovation Development Report and the Mianyang Initiative, while three national AI pilot bases (Mianyang, Qingdao, Xiamen) signed a model-data collaboration network agreement. The contest pioneered an "AI-evaluates-AI" approach — purpose-built assessment tools verifying agents' technical capability and runtime performance — and added its first international track.
From a ministry-run contest to an industry-wide model release, transportation is becoming one of the fastest industries for LLM adoption.
For full-stack AI platform vendors like NineZenith, "1,000+ works" means customers have crossed the awareness threshold: competition now shifts to delivery quality and iteration speed of industry models like TianXing in real road-network, logistics and safety scenarios.
(Compiled from public reports by China Transport News, Cover News and China Transportation News)