[DigitalToday reporter Jinju Hong (홍진주)] Alibaba has introduced a paid subscription model for its workplace artificial intelligence (AI) assistant QwenWork, stepping up efforts to monetise its AI services. The move aims to test actual demand for paid workplace AI agents after attracting users through free services.

The Hong Kong-based South China Morning Post (SCMP) reported on Aug. 11 that Alibaba's Qwen app is offering a three-tier membership for QwenWork users on Apple's App Store.

Annual subscriptions start at 200 yuan (about 40,000 won) for the cheapest plan. The top-tier version is priced at 1,499 yuan a year (about 310,000 won). QwenWork is a productivity-focused AI agent for workplaces that Alibaba began a public test of earlier this month.

It was offered for free during the public test to expand its user base. Alibaba has now added paid memberships, moving to a stage of testing the service's profitability.

By contrast, the general Qwen AI chatbot will remain free to use. Alibaba has chosen a strategy of setting a separate price for agent functions that carry out actual work, rather than making all its AI services paid.

The monetisation of QwenWork is closely tied to AI infrastructure costs, beyond simply putting a price tag on new features. Providing large language models (LLMs) and multimodal AI agents to many users requires substantial computing resources. In particular, AI agents that perform multi-step tasks beyond simple question-and-answer functions can see inference costs rise as usage increases.

As service users increase, AI infrastructure costs also rise. This appears to have been reflected in the decision that free services alone would make it difficult to cover long-term operating costs.

Alibaba is expanding Qwen beyond a simple chatbot into an AI agent ecosystem. The company said the Qwen open platform was designed so third-party developers and service providers can link various tools across devices including smartphones, personal computers (PCs) and smart glasses. It is also expanding use cases to more than 10 industries including logistics, real estate, finance and automobiles.

Qwen aims to handle real tasks that typically require users to go through multiple steps, through a conversational interface. For example, it can be instructed to check the location of a parcel delivered by SF Express, or to pass budget and location conditions to the real estate platform ZhuRong to find suitable listings. The approach is for AI to interpret user requests and connect with external services to carry out actual tasks, rather than simply providing information.

Additional charges linked to video generation functions have also been introduced. Depending on usage limits, the price of video generation credits was set from 26 yuan (about 5,446 won) to a maximum of 968 yuan (about 200,000 won).

The market is watching whether Alibaba can expand AI agent users while also securing paying customers. Wang Zhebin (왕쩌빈), an investment manager at Shenzhen-based JG Investment, assessed that monetising AI services and expanding users can proceed alongside technology improvements. He said that as users grow, feedback from actual use increases, allowing the AI system to be refined further.

Alibaba's strategy also aligns with a monetisation trend in the global AI industry. OpenAI and Anthropic are already operating paid subscription models built around premium features, and in China, Moonshot AI and Z.ai are also rolling out paid plans centred on advanced AI tools.

Against this backdrop, Alibaba is validating Qwen's business viability by first securing users through free services and then charging for workplace AI agents. Expectations have emerged that AI agents may offer stronger incentives to pay because they directly intervene in users' work processes more than existing chatbots. But a challenge remains that if they fail to provide sufficient performance and stability in real work, it will be difficult to raise the conversion rate to paid plans.

Ultimately, the success or failure of monetising QwenWork is expected to hinge on whether individual users and corporate customers will actually pay for the AI agent's work processing capabilities.

How much of the user base Alibaba has secured through free offerings it can convert into paying customers, and how much subscription revenue can offset rapidly rising AI infrastructure costs, is expected to be an important turning point for the future expansion of the Qwen ecosystem.