Doubao tests paid version, three-tier subscription revealed, AI enters the "tiered pricing" era.

Doubao tests paid version, three-tier subscription revealed, AI enters the "tiered pricing" era.

ByteDance's AI product Doubao has officially taken a key step toward commercialization. Its App Store page recently displayed a notice about paid version services, revealing three subscription tiers with the highest priced at 500 yuan per month. This development has sparked widespread discussion about whether the "free era" for China's large AI models is coming to an end. On May 4th, according to reports from Shanghai Securities News and The Paper, the Doubao App Store page now includes a paid version service notice, with three subscription tiers priced at 68 yuan per month for the standard version, 200 yuan per month for the enhanced version, and 500 yuan per month for the professional version. Doubao’s official response stated that the details of these plans are still in the testing phase, and full information will be released through official channels once it launches, while the free services will remain available. The launch of the paid versions signifies that Doubao is establishing a business model of “basic tools for free + professional scenarios for a fee.” Analysis points out that this shift demonstrates ByteDance is clarifying its monetization path in the large model sector and reflects how China’s AI industry, after amassing a large user base, is facing dual pressures of computing power costs and commercial sustainability. Three-tier pricing revealed, paid features focus on productivity scenarios According to the Doubao App Store page, the paid version comes in three price levels: Standard plan: continuous monthly subscription at 68 yuan per month (annual subscription at 688 yuan); Enhanced plan: continuous monthly subscription at 200 yuan per month (annual subscription at 2048 yuan); Professional plan: continuous monthly subscription at 500 yuan per month (annual subscription at 5088 yuan). Currently, Doubao’s app does not have related paid options or feature entry points. Doubao's official statement says, "The details of the plan are still under testing, and full information will be officially announced when it goes live." Reports cite insiders familiar with Doubao as saying that paid features will mainly focus on complex tasks and productivity scenarios, specifically including advanced requirements such as automatic PPT generation, deep data analysis, and video content creation. According to sources, as the model's capabilities continue to improve, the product can now meet increasingly complex, high-value tasks, but such tasks require more computing power and inference time, which is the core driver behind the rollout of paid services. The free version will continue to be available for everyday user needs. Computing power cost pressures drive business model transformation Doubao’s exploration of paid services is driven by the pressure of computing power costs due to rapid growth in large models. According to Volcano Engine, as of March this year, Doubao’s daily token usage has exceeded 120 trillion, doubling over the past three months and increasing 1000-fold since the release in May 2024. The number of enterprises on Volcano Engine with cumulative token usage exceeding one trillion has risen from 100 at the end of last year to 140. ByteDance Volcano Engine lead Tan Dai previously stated, “The price difference of tokens essentially reflects their capability difference.” He indicated that next-generation models have stronger capabilities, and the per-token cost will rise accordingly, creating greater economic value: "The increase in price with improved model intelligence fundamentally means it can create more value for clients." This rationale underpins Doubao’s logic for paid services—when the model's capabilities are sufficient to support high-value productivity tasks, charging professional users is reasonable and helps balance widespread adoption against computing power costs. Domestic and international AI paid landscape: Doubao is not the pioneer Globally, paid subscription models for AI products have long been mainstream. OpenAI's ChatGPT has multiple tiers—Free, Go, Plus, Business, Enterprise—with paid packages charged per month/user, and higher-priced offerings for teams and enterprises. Overseas leading models such as Anthropic and Grok have also adopted paid subscriptions. In China, leading vendors such as Zhipu and DeepSeek have already launched paid versions. Doubao’s move means mainstream large AI products in China have basically completed their paid services layout. It’s worth noting that domestic large models have not formed a unified pricing strategy. Zhipu has raised its API price three times this year: On February 12th, prices for the GLM Coding Plan package were adjusted, with overall increases starting at 30%; On March 16th, GLM-5-Turbo API prices were raised by 20%; On April 8th, GLM-5.1 was launched with another 10% price increase. After the adjustment, GLM-5.1’s token price for cache hits in coding scenarios is close to Anthropic’s Claude Sonnet 4.6 level. Meanwhile, DeepSeek is heading in the opposite direction. On April 25th, DeepSeek announced a limited-time 75% discount for V4-Pro model API; on April 26th, DeepSeek further announced that all API services’ input cache hit prices were lowered to one-tenth of their original prices. After the latest adjustment, the price for DeepSeek-V4-Flash input cache hit per one million tokens is 0.02 yuan, DeepSeek-V4-Pro is 0.025 yuan. This shows that whether prices go up or down, it is more a result of each company’s independent decisions based on its market positioning, and no collective industry trend has yet emerged. Analysts believe Doubao’s exploration of paid services is an autonomous choice seeking balance between user scale and commercial monetization, rather than a signal for industry-wide changes. Risk Warning and Disclaimer The market carries risks; investors must remain cautious. This article does not constitute personal investment advice, nor does it take into account the special investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article fit their specific circumstances. All risks associated with investment are borne by the investor.