Doubao’s Paid Plans

Recently, the topic of “Doubao Paid” has gained significant attention as Doubao announced its transition to a paid model. According to the service statement updated on the App Store, users will be categorized into four tiers:

• Basic Version: Remains free for daily needs;
• Standard Version: Monthly subscription at 68 yuan, annual subscription at 688 yuan;
• Enhanced Version: Monthly subscription at 200 yuan, annual subscription at 2048 yuan;
• Professional Version: Monthly subscription at 500 yuan, annual subscription at 5088 yuan.

Insiders indicate that the paid features will focus on complex tasks and productivity scenarios, such as PPT generation, data analysis, and film production.

ByteDance’s Financial Strategy

From the subscription fee structure, it is clear that Doubao has put considerable thought into its pricing: the Standard Version at 68 yuan is less than half of ChatGPT Plus’s monthly fee of 142 yuan. The Professional Version’s annual fee of 5088 yuan exceeds what individual users typically pay, directly targeting enterprise-level services.

This indicates that ByteDance aims to showcase its “cost-effectiveness” while accurately targeting different user segments to enhance conversion rates. Currently, Doubao has not opened payment options across its product lines and has not clarified the differences in rights among the various versions. The official response states that Doubao will explore additional value-added services while continuing to provide free services, with details still in the testing phase.

Despite ByteDance’s cautious approach, the decision to charge for Doubao has sparked significant interest among users and industry observers, becoming a polarizing topic. Concerns have been raised about the potential reduction of features in the free version, with some humorously suggesting that while users thought they were using Doubao for free, they were actually helping to improve its model by providing data.

Financial Pressures and AI Investment

Recently, media reports indicated that ByteDance’s net profit for 2025 could decline by over 70%. However, Douyin’s Vice President Li Liang clarified that this figure was based on international accounting standards, and the actual operating profit margin had only “slightly decreased,” far less than reported. He acknowledged, however, that the slowdown in Douyin e-commerce growth and increased investment in emerging businesses would indeed pressure the operating profit margin in the second half of the year.

Li’s response is intriguing as he asserts that ByteDance’s operational health remains intact, with growth in Douyin e-commerce and other businesses. The “slight decrease” is attributed to accounting measures.

The real issue lies in the increased investment in emerging businesses, particularly AI, which includes rising costs for computing power, model development, and talent recruitment. In early 2026, ByteDance’s CEO Liang Rubo defined AI as a “core opportunity larger than PC and Web combined,” emphasizing that the immediate goal is to scale Doubao and its ToB (Model as a Service) business.

This directive shapes ByteDance’s resource allocation logic for 2025 and beyond: profit is secondary to securing a foothold in the AI era. This creates an internal conflict within ByteDance, balancing traditional internet traffic businesses—primarily selling ad space, which generates substantial revenue—against AI and other emerging businesses that, despite having millions of users, still rely on advertising for funding.

ByteDance’s capital expenditure for 2025 is expected to exceed 150 billion yuan, with about 90 billion yuan allocated for AI computing power procurement. This figure is projected to rise to 160 billion yuan in 2026, with 85 billion yuan for AI chip procurement. AI has become a core area for ByteDance’s capital investment.

The Challenge of Monetization

The pressing question remains: who will pay for Doubao? ByteDance’s urgency to address the common challenge of user growth outpacing the establishment of a viable business model is evident.

Doubao’s introduction of a three-tier monthly subscription system reflects a strategy to keep basic functions free to maintain a broad user base while creating a paid channel for heavy productivity scenarios, validating the monetization model through user segmentation.

This commercial framework is mature and closely mirrors the value-added payment models of leading overseas firms like OpenAI and Anthropic. In contrast, previous attempts by domestic AI companies to monetize have often remained in the experimental phase, cautious and conservative.

For example, Baidu’s Wenxin Yiyan briefly launched a membership system but retracted it due to public backlash. Other emerging AI players like Kimi, Zhipu Qingyan, and iFlytek’s Spark have offered low-priced services with vague functionality distinctions, catering only to niche professional users without engaging the broader market.

In comparison, Doubao’s approach appears aggressive and Westernized, attempting to establish a mindset that “payment equals professional productivity” by categorizing high-demand scenarios as paid services.

However, the user base of Doubao complicates this strategy. Many users come from the Douyin ecosystem or are pre-installed on devices, with a significant proportion being from lower-tier markets, older adults, or light entertainment users. According to a QuestMobile report, the AI industry user growth is trending towards both the elderly and lower-tier markets in early 2026.

These users’ core needs can be met by the free version, but they may have concerns about the speed and accuracy of the free version. The paid model primarily serves professionals, creators, and developers, which may limit large-scale conversion.

More critically, competitors like Alibaba’s Qianwen, Tencent’s Yuanbao, and Baidu’s Wenxin, which offer comparable model capabilities and product experiences, may respond to Doubao’s full rollout of the paid system by introducing free high-level features or low-cost packages, thereby siphoning users and diluting Doubao’s paid value.

In my view, Doubao’s current signal of introducing paid features holds more strategic value than immediate revenue potential. Given its current conversion rates, Doubao’s revenue cannot cover the billions spent on computing power.

Expecting Zhang Yiming to achieve self-sufficiency in the short term is more challenging than creating another Doubao.

ByteDance itself lacks significant confidence in fully monetizing Doubao. Its core aim is to use this public paid testing to gauge three critical industry metrics: the payment threshold for domestic users, the upper limit of AI productivity service value, and competitors’ strategies in response to commercialization.

This move breaks the silence in the domestic AI industry, where the willingness of Chinese users to pay for AI has long been an open question. Most companies have opted for a conservative wait-and-see approach, hesitant to break free from the comfort zone of free services.

ByteDance’s actions have shattered this silence. The current exploration of paid services is not primarily about profit but about experimentation, breaking new ground, and establishing industry norms to end the embarrassing reality of the AI industry, which has been burning cash without generating revenue.

Ultimately, whether Doubao can sell itself as a valuable asset is becoming less significant.