China AI Native Industry Insights – 20260622 – Kunlun Tech | Alibaba | Kimi | more
Explore Skywork 3.1, LOGOS foundation model, Kimi Work updates, Qoder enhancements. Discover more in Today’s China AI Native Industry Insights.
1. Skywork releases version 3.1 with Skywork Design canvas interface and Dynamic Workflows multi-agent orchestration
Skywork has launched version 3.1 featuring two new capabilities: Skywork Design, a canvas-based interface for iterative UI/UX design work, and Dynamic Workflows, which orchestrates dozens to hundreds of parallel agents for large-scale tasks. Skywork Design addresses the need for complex, multi-page design projects that require 30-40 rounds of iteration, allowing users to manage entire UI systems with consistent design standards across pages. Dynamic Workflows enables automatic task decomposition and parallel execution with cross-verification, suited for batch operations like generating hundreds of product pages or site-wide content revisions. Both features are integrated with the main Tiangong platform and support export to Figma, web deployment, and developer handoff formats.
Read more: https://mp.weixin.qq.com/s/BshpodcEwshDfKd7PWjzMQ
Video Credit: The original article
2. LOGOS unified scientific foundation model open-sourced by ATH-Token Foundry and Renmin University
ATH-Token Foundry and Renmin University of China’s Gaoling School of Artificial Intelligence have open-sourced LOGOS, a multi-domain scientific generative foundation model based on a unified scientific grammar. The model uses a sequence-based paradigm to handle proteins, small molecules, materials, and chemical reactions within a single framework aligned with large language model architecture. LOGOS-1B demonstrates high parameter efficiency, outperforming larger models like NatureLM (8×7B) with only 1/56 of the parameters across six scientific tasks including pocket-conditioned ligand generation, retrosynthesis prediction, and MOF material generation. The model weights, inference code, and technical report are fully open-sourced on HuggingFace and GitHub.
Read more: https://mp.weixin.qq.com/s/50q5uY849FKnBzk1Q04MRg
Video Credit: NotebookLM
3. Kimi Work updates with Goal mode for 24-hour continuous tasks, Plugin Center, and June promotional discount
Moonshot AI announced updates to Kimi Work Beta version, including a Goal mode that enables continuous autonomous task execution for up to 24 hours using a loop mechanism, and a Plugin Center that integrates with third-party applications such as Baidu Netdisk, DingTalk, Feishu, WPS, and Canva. The Goal mode is designed for tasks with clear objectives but uncertain paths, allowing the AI agent to autonomously iterate toward completion while remaining transparent and interruptible by users. As a promotional offer for June 2026, all Kimi Work tasks consume member quota at a 50% discount when used through the desktop client.
Read more: https://mp.weixin.qq.com/s/KJav-s9qlkzV9yN8r6-sNg
Video Credit: The original article
4. Qoder Desktop adds Computer Use support for Windows
Qoder has released Windows support for its Computer Use feature in Qoder Desktop version 1.9.1. The feature allows users to describe tasks in natural language using the /computer-use command, and Qoder will autonomously control the Windows desktop to complete them, including launching applications, clicking interface elements, entering text, and handling popups. Unlike the Mac version, operations occur in the foreground with visible cursor movement, and the system can detect and handle Windows dialog boxes. Use cases include code verification by running applications, extracting information from GUI tools to generate code, troubleshooting multi-step issues, automating repetitive tasks in applications like Excel, and performing unfamiliar system configurations.
Read more: https://mp.weixin.qq.com/s/s6-jh-D3u70zOeyyjm2nsw
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.