Alibaba Qwen

Qwen, known in Chinese as Tongyi Qianwen (通义千问) and branded internationally as Qwen, is Alibaba Group's flagship family of large language and multimodal artificial-intelligence models. It was introduced in 2023 by Alibaba's cloud and research organizations as Alibaba's answer to the global generative-AI wave, and it sits inside the Alibaba Cloud Intelligence business, which commercializes the models through APIs, enterprise solutions and its model studio platform. The brand covers a broad model line rather than a single product: general-purpose chat and instruction-tuned language models, coding-oriented models, mathematics and reasoning variants, vision-language models, audio and speech models, and smaller open-weight models intended for local and edge deployment. Qwen models are released under a dual strategy — proprietary hosted endpoints on Alibaba Cloud plus openly downloadable weights for many sizes — which has made the Qwen family one of the most widely used open model lineages among developers worldwide. Commercially, Qwen functions as a platform brand: developers access it through Alibaba Cloud's model service, enterprises integrate it into applications, and Alibaba positions it as infrastructure for the Chinese and international AI ecosystem. Its consumer-facing presence is through Alibaba's own assistant applications and integrations across the group's products. Ownership is entirely within Alibaba Group, a public company listed in New York and Hong Kong. Qwen is not a separately listed entity; it is a strategic product brand within Alibaba's cloud and AI portfolio, and its releases are closely followed as indicators of Alibaba's competitive standing in the global AI race.

Qwen emerged from Alibaba's decision to build and commercialize its own foundation models as generative AI accelerated globally in 2023. Alibaba, whose cloud arm had long invested in AI research through its DAMO Academy, consolidated that work into a model family it announced in 2023 under the Chinese name Tongyi Qianwen (通义千问). The launch was framed as Alibaba's entry into the large-model race and was tied to Alibaba Cloud's strategy of selling AI compute and model services to enterprises. A defining strategic choice was openness. Alongside hosted API access, Alibaba began releasing Qwen model weights publicly, starting with smaller and mid-sized models and expanding to larger and more specialized variants. This open-weight approach lowered adoption barriers for developers, researchers and startups, and it helped the Qwen name gain recognition well beyond Alibaba's existing cloud customer base. Over successive releases the family broadened from text-only chat models to instruction-tuned models, coding models, mathematics and reasoning models, vision-language models and audio models, with multiple parameter sizes offered for different deployment contexts. The brand also became a consumer-facing name inside Alibaba's ecosystem, appearing in the group's assistant applications and integrated into Alibaba's various products and services. Internationally, Alibaba marketed the family under the shorter Qwen name, aligning the Chinese and global identities of the brand. As competition in generative AI intensified among Chinese and international technology companies, Qwen releases became regular news events, with each new generation watched for benchmark performance, openness and pricing. Alibaba positioned Qwen as core infrastructure for its cloud business, using the models to attract developers and enterprise workloads to its platform. The brand thus evolved from a single 2023 launch into a sustained, multi-generation model program that spans research, open-source community engagement and commercial cloud services, all under Alibaba Group ownership.

• Alibaba announces the Tongyi Qianwen (Qwen) large model family in 2023.

• Alibaba expands the family beyond text chat into coding, math, vision-language and audio models across successive releases.

• Alibaba releases open-weight Qwen models alongside hosted API access, building a large global developer following.

Alibaba's flagship generative-AI model family, positioned as both a cloud-hosted enterprise platform and a leading open-weight model lineage for global developers.

• Qwen general-purpose language models

  Instruction-tuned chat and general-purpose text models offered in multiple sizes.

• Qwen coding models

  Models tuned for code generation and software development tasks.

• Qwen vision-language models

  Multimodal models handling image and text inputs.

• Qwen audio models

  Speech and audio-oriented models within the family.

• Alibaba Cloud model service

  Hosted API access to Qwen models for developers and enterprises.