Skip to main content

Qwen (Alibaba Cloud)

Category: Foundation Models / LLMs

Alibaba Cloud's foundation model and AI business unit, developing the open-weight Qwen family of LLMs, multimodal models, and voice/agent products. Qwen (Alibaba Cloud) was founded in 2023. Based in Hangzhou, China. Latest round: Undisclosed.

Founded
2023
Headquarters
Hangzhou, China

Value proposition

Provides a full-stack, open-source-leaning ecosystem of large language, multimodal, and voice models deployable both via self-hosted open weights and Alibaba Cloud's managed Model Studio API, powering consumer apps and enterprise AI adoption at scale.

Products and solutions

Qwen3 / Qwen3.5 / Qwen3.6 / Qwen3.7 / Qwen3.8-Max LLM series (flagship 2.4T MoE), Qwen-Audio / Qwen3-TTS, CosyVoice Studio (CosyFlow, CosyAgent, CosyCreative), Qwen3-ASR, Qwen-VL multimodal models, Qwen-Image / Qwen-Image-3.0, Qwen-Robot Suite (RobotNav, RobotManip, RobotWorld), Qwen Code, Qwen App / Qwen Studio (consumer assistant), Alibaba Cloud Model Studio API / QwenCloud

Unique value

Largest and most prolific open-weight LLM ecosystem from China with nearly 1B cumulative downloads and >50% of global open-source model download share, combining open-source community traction with Alibaba Cloud's commercial deployment infrastructure

Target customer

Enterprises and developers in China and globally (via Alibaba Cloud), consumer end-users of Qwen app/DingTalk, AI researchers/open-source community

Industries served

Cross-industry: enterprise SaaS, cloud computing, consumer apps, developer tools, robotics/physical AI (via NVIDIA integration)

Technology advantage

Deep integration with Alibaba Cloud infrastructure, large-scale proprietary training data and compute, rapid iterative releases through Qwen3.8-Max (2.4T MoE, first Max-class open-weight commitment), voice stack (Qwen-Audio / CosyVoice Studio), and embodied/physical AI (Qwen-Robot Suite + NVIDIA Physical AI stack)

How they differentiate

Broadest open-weight model breadth and cloud deployment ecosystem among Chinese AI labs; smoothest international developer onboarding (OpenAI SDK-compatible, English docs, global billing) versus more domestically-focused rivals like Doubao/ERNIE

Main competitors

DeepSeek, ByteDance Doubao, Moonshot AI (Kimi), Zhipu AI (GLM), Baidu (ERNIE)

Key partnerships

NVIDIA (physical AI software stack integration for robotics/simulation on Alibaba Cloud), Hugging Face and ModelScope (open model distribution), DingTalk (enterprise deployment)

Notable customers

90,000+ enterprises via Alibaba Cloud & DingTalk deployment, NTUC Union Companies (Singapore, enterprise AI training pilot)

Major milestones

April 2023: Tongyi Qianwen (Qwen) launched, Rapid open/proprietary releases through Qwen3, 3.5, 3.6, 3.7, March 2026: tech lead Junyang Lin and other senior researchers depart; Alibaba forms AI task force, June 2026: Tongyi + Future Life Lab merged into Token Foundry under CEO Eddie Wu; Zhou Jingren becomes Alibaba Chief Scientist, Aug 2–3 2026: Qwen3.8-Max released (2.4T MoE, open weights planned), Aug 7 2026: CosyVoice Studio voice AI platform launched (CosyFlow, CosyAgent, CosyCreative on Qwen-Audio)

Growth metrics

Nearly 1B cumulative open-model downloads by March 2026 (>50% of global open-source model downloads per Interconnects/SCMP); Qwen App 100M+ MAU by Jan 2026 (SCMP), with later reports of 100M–200M+ scale; 90,000+ enterprises via Alibaba Cloud & DingTalk

Market positioning

Leading open-source LLM ecosystem play in China, positioned as the 'open weights + enterprise cloud' alternative to closed-model competitors, with voice AI (CosyVoice Studio) as a new productivity frontier

Geographic focus

China domestic market primary; global expansion via open-weight downloads and Alibaba Cloud international enterprise customers

Latest news about Qwen (Alibaba Cloud)

More Foundation Models / LLMs companies

Official website: