Eastmoney analysis links China’s model upgrades to agent commercialisation

From model upgrades to workflows

Eastmoney Research Center’s October 20 analysis argues that competition among Chinese large language models is shifting toward commercial deployment of AI agents. It links MiniMax’s October 19 M3.1 Flash release and DeepSeek’s earlier model and search-interface upgrades with demand for systems that can break down tasks, call tools and deliver results.

The report cites Chinese brokerages’ view that more tasks per user could increase token consumption and recurring revenue. It attributes to Orient Securities the argument that competition is moving toward defining specialised tasks, accumulating real workflow records and building feedback mechanisms. These are the analysts’ expectations, rather than independently established revenue outcomes.

A broad listed-company ecosystem

Eastmoney also provides its own market snapshot as reported in the article: its agent-themed A-share category contains 183 companies with a combined market value of about 2.24 trillion yuan. Its Choice data identifies 17 companies with net margin-financing purchases since the start of October. The figures describe Eastmoney’s category and investor activity at the report’s stated time, rather than the entire market. They do not establish how much revenue those companies earn from agents.