DeepThink and MiniMax Obtain Approval for AI Model Export
October 26, 2026 · 2 min read
DeepThink and MiniMax, companies from China, have completed the export registration procedures for the overseas release and licensing of their open-source AI models. This development comes as the Chinese government is reportedly pushing for tighter controls on the export of AI technology while still allowing domestic companies to pursue overseas business and the commercialization of AI models.
On October 22, the Chinese state-run CCTV-affiliated social media account, Yuyuantian, announced that DeepThink and MiniMax received approval for the overseas use of their AI models. MiniMax also confirmed on the same day that it had completed the technical export registration with the Ministry of Commerce of China. The registration pertains to certain models that have recently been released or will be made available in the future, facilitating both open-source release and international distribution.
Yuyuantian highlighted in its post that following the policy announcement, DeepThink and MiniMax successfully completed the registration and related procedures for open-source release and overseas usage permissions. The post further stated that AI companies can continue their commercialization activities as usual.
This registration occurs amidst the Chinese government's efforts to enhance export controls on AI technology. On October 16, the Ministry of Commerce proposed export control measures covering 23 areas of AI technology, including models, AI agents, and computing clusters. The intention is to restrict the overseas transfer of key technologies deemed critical to national security.
However, the Chinese government maintains that most AI technologies can still be freely exported and that simplified approval processes will apply to the export of open-source technologies and the operation of overseas data centers by Chinese companies. Yuyuantian emphasized that the revision of the export control list is not intended to block the overseas transfer of AI technology entirely, but rather to establish clear boundaries on what is permitted.