en.Wedoany.com Reported - The daily token usage of Doubao has exceeded 120 trillion, a 1000-fold increase compared to its initial release in May 2024. According to a disclosure by Volcano Engine on April 2nd, by March 2026, the daily token usage of the Doubao large model had surpassed 120 trillion, with this number doubling over the past three months. When Doubao was first publicly released in May 2024, its daily token usage was at a level of 120 billion. As explained by Tan Dai, President of Volcano Engine, Doubao's daily token consumption was approximately 63 trillion by the end of 2025. The number of enterprise customers with cumulative token usage exceeding one trillion has grown from 100 at the end of 2024 to 140. Tan Dai pointed out that token usage is a core metric for measuring the development speed of AI, and the recent growth is primarily driven by the explosion of AI video creation and the accelerated adoption of AI agents.
Volcano Engine's MaaS (Model as a Service) business revenue target for 2026 is over 10 billion yuan. According to Tan Dai, the team had previously set an annual revenue target of 100 billion yuan in their 2021 business plan. In 2025, this target was proactively raised by hundreds of billions of yuan, while the original achievement timeline of 2029-2031 remains unchanged. Regarding the large model pricing strategy, Tan Dai stated that the pricing for each generation of Doubao is meticulously designed and will not change after being set. Doubao 2.0 Pro is priced at 3.2 yuan per million tokens for input (within 32k context) and 16 yuan per million tokens for output. Tan Dai believes that token price cannot be discussed in isolation from model value. The next-generation model is more capable, and the economic value per token is higher; theoretically, the price should increase.
On the same day, the Seedance 2.0 API officially entered public beta for enterprise users. According to the Volcano Engine announcement, this model adopts a multimodal audio-video joint generation architecture, supporting text, image, audio, and video as input modalities. It enables character consistency, camera control, physical simulation, and native audio-visual synchronization generation. Target scenarios include advertising and marketing, content production, and short video creation. Volcano Engine has established a comprehensive copyright and portrait security system covering the entire process for this model.
The ArkClaw agent service continues to iterate. This product is a cloud-based SaaS version agent platform developed based on the open-source OpenClaw and provided by Volcano Engine. Users can access a 7×24 online AI assistant without local deployment, which can connect to Feishu and integrate with messaging channels like WeChat, DingTalk, and Weibo. According to Tan Dai's analysis, the core elements for enterprise agent applications include model performance, security mechanisms, and skills. In terms of security, vendors need to provide well-configured clauses in a productized manner; regarding skills, maintaining an open ecosystem is essential. Volcano Engine has collaborated with OpenClaw to establish the ClawHub China mirror site, providing Chinese developers with stable and fast access to the skill library.
Tan Dai stated that not everyone needs to use OpenClaw-like agents; most ordinary users can meet their needs by using Doubao. However, the audience for agents might constitute 20% to 30% of the population proportionally. Regarding enterprise Agent demand, Tan Dai believes the industry will develop along a dual-path approach: Agile Agents, represented by products like ArkClaw, focus on enhancing individual productivity; Stable Agents help enterprises achieve process-oriented and large-scale AI implementation. These two paths complement each other to drive enterprise AI transformation and upgrading.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
