Alibaba Cloud used its first international Qwen Conference in Singapore on 26 May 2026 to unveil a full‑stack agentic AI ecosystem for global customers, combining new models, platform features, infrastructure upgrades and agent products. The package is positioned to accelerate the “agentic” era by enabling AI agents to call cloud services directly, reducing the engineering friction that has limited real‑world deployment of multi‑step workflows.
The company released its latest large language model, Qwen3.7 — Max, to Model Studio in the Singapore region. Citing the Artificial Analysis global large language model Intelligence Index presented at the event, Alibaba Cloud said Qwen3.7 — Max ranked fifth globally and first among Chinese models with a score of 56.6, outperforming Kimi — K2.6, DeepSeek — v4‑Pro‑Max and GLM5.1 while remaining competitive with international offerings such as GPT, Claude and Gemini.
To let agents operate across cloud services, Alibaba Cloud launched a Skills portal that converts common capabilities from more than 60 cloud products into Skill‑based, MCP‑compatible formats so agents can invoke cloud resources similar to calling functions. The company also outlined product‑level agents for core offerings — databases, big data, operations and maintenance (O&M), and security — intended to simplify management and reduce the operational load of complex cloud environments.
On the infrastructure side, Alibaba Cloud said it is upgrading runtime support for agents with several technical enhancements: lightweight execution sandboxes to isolate tasks, cross‑task memory to preserve state between steps, seamless data circulation between components, and intelligent O&M across the full technology stack. These changes are aimed at lowering friction for agents executing multi‑step workflows and improving integration with cloud services at scale.
Alibaba Cloud also introduced Qwen Cloud, an AI‑native cloud platform designed to centralize model serving and agent access. The platform uses a three‑entry design — “Skills” for agent access, a command line interface (CLI) for workflow integration, and a user‑facing website — and aggregates proprietary Qwen models, open‑source models and third‑party offerings across text, vision, audio, image, video and embedding tasks to support developers, enterprises and prosumers.
Reinforcing local adoption, the company announced a collaboration with the Tech Talent Assembly (affiliated with the National Trade Union Congress) and ST Telemedia Global Data Centres (STT GDC) to deliver practical generative and agentic AI skills to more than 1,000 SMEs and students in Singapore. The initiative, announced at the Qwen Conference, aims to widen the local talent pipeline and accelerate real‑world usage of the company’s agentic tooling.
Sources
Replies (0)
No replies in this topic yet.
