China AI Native Industry Insights – 20260818 – Z.ai | Alibaba | Kunlun Tech | more
Today’s digest highlights key developments from Z.ai and Alibaba Wan Video, illustrating advancements in AI capabilities and creative workflows. A prominent theme connecting these stories is the enhancement of specialized AI systems for improved functionality and creative expression. Z.ai’s GLM-5.3 demonstrates notable improvements in coding and cybersecurity capabilities via post-training techniques, while Alibaba Wan Video’s launch of Agent Skills offers community-driven creative workflows. Additionally, Kunlun Wanwei’s Mureka V9.5 AI model introduces better natural sound with enhanced Chinese vocal performance, signifying notable strides in AI-generated music quality. Discover more in Today’s China AI Native Industry Insights.
1. Z.ai Releases GLM-5.3 with Improved Coding and Cybersecurity Capabilities via Post-Training
Z.ai released GLM-5.3 on August 14, 2026, building on the same 743B base model as GLM-5.2 with all capability gains derived from scaled-up post-training. The model targets coding and agentic tasks, with Terminal-Bench 3.0 scores rising from 4.6 to 28.3, positioning it as a top performer among open-weight coding models. GLM-5.3 also introduces notable advances in cybersecurity, which Z.ai says exceeded expectations relative to training investment. The model is available through Z.ai’s GLM Coding Plan, with open weights planned for release approximately two weeks after launch.
Read more: https://z.ai/blog/glm-5.3
Video Credit: NotebookLM
2. Alibaba Wan Video Launches Agent Skills for Community-Built Creative Workflows
Alibaba’s Wan video platform has launched Agent Skills, a new feature that allows users to build and share custom AI-powered creative workflows. Community members can explore Skills created by others or publish their own, with examples including Short Drama Generators and Music Video Generators. The feature is designed to turn user ideas into reusable creative workflows that can be shared publicly. To encourage adoption, Alibaba is running a reward campaign through August 31, offering 150 Inspiration Points for each successfully published Skill, with a cap of 450 points per account.
Read more: https://x.com/Alibaba_Wan/status/2088092769709690951
Video Credit: @Alibaba_Wan on X
3. Kunlun Wanwei releases Mureka V9.5 AI music generation model with improved natural sound and Chinese vocal performance
Kunlun Wanwei announced the release of Mureka V9.5, a new AI music generation model built on the MusiCoT (Music Chain-of-Thought) framework. The update focuses on more natural and restrained music arrangement rather than dense, overly complex compositions, achieving a 61.0% vocal quality rate, 97.0% control quality rate, and 95.7% style accuracy. The model shows significant improvements in Chinese-language music generation, including better pronunciation accuracy, more natural vocal delivery, improved harmony arrangement, and authentic instrumentation for traditional Chinese music styles. The development incorporated feedback from over 25,000 users globally and is now available on both international and domestic platforms.
Video Credit: NotebookLM
4. Qwen Office open-sources MyContext, a context infrastructure project for workplace AI agents
Qwen Office has open-sourced MyContext, a context infrastructure project designed to provide AI agents with workplace context from multiple data sources. The project processes data from instant messaging platforms like DingTalk and Feishu, documents, meetings, and collaboration records into a continuously updated work profile that agents can query and reason with. MyContext runs locally on user devices and uses semantic analysis to handle conflicting information, allowing users to build personal AI assistants that can automatically respond to work messages. The project has gained over 1,000 GitHub stars within one week of release.
Video Credit: NotebookLM
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.