China AI Native Industry Insights – 20260908 – OpenBMB | Tencent | Qoder | more
Today’s digest highlights significant developments from OpenBMB, Tencent, and Qoder as they make strides in AI technology. A common theme is enhancing AI model capabilities and efficiency, evident from OpenBMB’s open-sourcing of the MiniCPM5-2B, a highly-ranked sub-4B parameter edge language model, and Tencent’s efficiency upgrade to its Hunyuan Preview Model. Additionally, Qoder’s new desktop application now supports mobile remote control, providing users with increased flexibility and convenience. These advancements reflect the ongoing push for improved performance and user accessibility in AI applications. Discover more in Today’s China AI Native Industry Insights.
1. OpenBMB Open-Sources MiniCPM5-2B, a Top-Ranked Sub-4B Edge Language Model
OpenBMB has released MiniCPM5-2B, a 2-billion-parameter open-source language model designed for edge deployment. The model ranks first among open-source models under 4B parameters on the Artificial Analysis Intelligence Index with a score of 23, and scores 20 on the Agentic Index. Across 34 benchmarks covering coding, math, long-context understanding, tool use, and agentic tasks, MiniCPM5-2B achieves an average score of 53.9. Alongside the model weights, OpenBMB is also releasing the training data, training recipes, and reinforcement learning stack used to build it.
Read more: https://www.openbmb.cn/model/minicpm5-2b
Video Credit: NotebookLM
2. Tencent Ships Efficiency Upgrade to Hunyuan Preview Model
Tencent’s Hunyuan team released an upgrade to the Hy4 preview model addressing user-reported issues of long thinking times and over-verification on complex tasks. The update maintains the same task quality while reducing the number of turns and lowering input and output token usage. Tencent says both benchmark testing and human evaluation confirmed the improvements. The upgraded model is now live for all users and can be tried on the WorkBuddy platform.
Read more: https://mp.weixin.qq.com/s/ROjTg1fmp-Ff4BqDJez2GQ
Video Credit: NotebookLM
3. Qoder releases desktop application with mobile remote control support
Qoder announced a new desktop application that enables users to initiate and manage coding tasks remotely from mobile devices. The update allows users to assign tasks via voice input, monitor multi-step workflows handled by sub-agents, and review outputs including code, documents, and HTML previews directly on mobile. Enterprise administrators can configure download and sharing permissions for security management. The mobile app connects to the desktop version to leverage local files and development environments while providing real-time task notifications and authorization requests.
Read more: http://mp.weixin.qq.com/s?__biz=MzE5ODI2NzI5Nw==&mid=2247491743&idx=1&sn=de7eb954aa45e6b18886d611906397cb&chksm=971d7826857492b56c7baac22d7c2ae83984c0920f52e7cf43f7ac33b0e8ac022cb1b0b510ac&scene=0&xtrack=1#rd
Video Credit: The original article
That’s all for today’s China AI Native Industry Insights. Join us at AI Native Foundation Membership Dashboard for the latest insights on AI Native, or follow our linkedin account at AI Native Foundation and our twitter account at AINativeF.