Alibaba Cloud's Qwen 3.7 Max runs autonomously 35 hours

At its inaugural Qwen Conference in Singapore, Alibaba Cloud unveiled Qwen 3.7-Max, an agentic AI model designed for autonomous, multi-step tasks. The model features over one trillion parameters and a one-million-token context window, enabling it to run for 35 hours without performance degradation. Ken Xu, solution architect director, highlighted its ranking among top models on the Artificial Analysis Intelligence Index. CTO Li Fei Fei declared the arrival of an 'agent workforce' that will multiply human productivity. The company also launched Qwen Cloud, a platform serving both humans and AI agents, and a skills portal converting cloud capabilities into agent-compatible formats. For enterprise deployments, the JVS Agent Suite provides secure toolkits with OpenClaw agents.

Alibaba Cloud addressed workforce displacement fears by partnering with Singapore's NTUC and ST Telemedia to equip over 1,000 local SMEs and students with genAI and agentic AI skills. Starting June 2026, eligible workers will receive tokens for Qwen, Wan video generation, and Qoder coding platform subscriptions. Senior Minister Desmond Tan emphasized a 'no jobless growth' approach. Additionally, Alibaba highlighted Qwen's role in regional projects like Sea-Lion, Southeast Asia's first LLM fine-tuned from Qwen architecture, now used by enterprises across ASEAN for complex translations and cultural adaptation.