Meituan releases open-source video AI model
Meituan on Oct. 27 released an open-source video generation model named LongCat-Video. The model has 13.6 billion parameters and is based on the Diffusion Transformer (DiT) architecture. Meituan said the model can stably generate videos up to five minutes long without quality loss, thanks to pre-training on video continuation tasks. It is designed for scenarios requiring long-term dynamic simulation, such as digital humans and embodied AI.
China’s securities regulator to deepen ChiNext reforms
Wu Qing, chairman of the China Securities Regulatory Commission (CSRC), said on Oct. 27 that the regulator will deepen reforms of the tech-focused ChiNext board. The plan involves setting listing standards better suited to the characteristics of innovative and entrepreneurial enterprises in emerging and future industries. The goal is to provide more precise and inclusive financial services for companies in new industries, new business formats, and new technologies.
WeRide reveals Hong Kong IPO plan
Autonomous driving firm WeRide Inc. announced plans for its Hong Kong public offering in a bourse filing on Oct. 27. The company plans to globally issue 88.25 million H-shares, subject to an overallotment option, at a maximum price of HK$35 ($4.50) per share. Pricing is expected on Nov. 4, with trading set to begin on Nov. 6.
XPeng expands into the Baltics and Cambodia
XPeng Inc. on Oct. 27 detailed its global expansion in October, entering the markets of the three Baltic states — Lithuania, Latvia, and Estonia — as well as Cambodia. In the Baltics, XPeng is introducing several models including the G6 and G9, and is collaborating with leading local dealers to build sales and after-sales networks. In Cambodia, the company launched six models at once, with deliveries scheduled to start in November. To date, XPeng’s sales and service network has expanded to over 49 countries and regions.
MiniMax launches new inference model
Chinese AI unicorn MiniMax on Oct. 27 launched a new open-source large inference model, M2. The model uses a Mixture-of-Experts (MoE) architecture with a total of 230 billion parameters, of which 10 billion are active. It boasts an output speed of 100 tokens per second. MiniMax said the model is specifically designed for developing AI agents.
China’s 5G base stations reach 4.7 million
China had a total of 4.705 million 5G base stations as of the end of September, an increase of 455,000 from the end of last year, according to data released by the Ministry of Industry and Information Technology on Oct. 27. 5G stations now account for 36.6% of all mobile base stations, up 0.9 percentage points from the first half of the year. In the first three quarters of 2025, telecom business revenue totaled 1.327 trillion yuan ($186.6 billion), a year-on-year increase of 0.9%.
