China AI Native Industry Insights – 20260421 – Moonshot AI | Alibaba | more
Explore Kimi K2.6, Fun-ASR1.5, and Qwen3.6-Max-Preview advancements. Discover more in Today’s China AI Native Industry Insights.
1. Moonshot AI releases Kimi K2.6 open-source coding model with 300-agent swarm capabilities
Moonshot AI has officially released Kimi K2.6, an open-source coding model that achieves state-of-the-art performance on multiple benchmarks including SWE-Bench Pro (58.6%) and SWE-bench Multilingual (76.7%). The model enables ultra-long coding tasks with over 4,000 tool calls and continuous execution for more than 12 hours. Key features include 300 parallel sub-agents per run, advanced frontend development with video backgrounds and 3D effects, and proactive agents for 24/7 autonomous operation. K2.6 is now available on kimi.com in chat and agent modes, with API access and open weights published.
Read more: https://mp.weixin.qq.com/s/6jfSSCcq7HMg-qXrsc4OVg
Video Credit: The original article
2. Alibaba Tongyi Lab launches Fun-ASR1.5 speech recognition model supporting 30 languages
Alibaba Tongyi Lab officially launched Fun-ASR1.5, an end-to-end speech recognition large model that supports 30 languages within a single unified architecture. The model features multilingual support across Asia, Europe, and the Middle East, native code-switching capabilities for mixed-language speech, and professional text output with smart punctuation and automatic formatting. Fun-ASR1.5 has been made available on Alibaba Cloud’s Bailian platform, providing API services to clients across education, media, finance, technology and culture industries.
Read more: https://mp.weixin.qq.com/s/xPEstUiJ4IC3BOFVsdvRLA
Video Credit: @Ali_TongyiLab on X
3. Alibaba releases Qwen3.6-Max-Preview model with improved coding and agentic capabilities
Alibaba released an early preview of Qwen3.6-Max-Preview on April 20, 2026, positioning it as the most powerful model in the Qwen series. The proprietary model demonstrates significant improvements in agentic coding capabilities, achieving top scores across six major programming benchmarks including SWE-bench Pro and Terminal-Bench 2.0. The model also shows enhanced world knowledge and instruction-following abilities compared to the previous Qwen3.6-Plus, with availability through Qwen Studio and Alibaba Cloud Model Studio API.
Read more: https://mp.weixin.qq.com/s/dIzM25Cn9kv-8KLWxwKzMA
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.