The Zhitong Finance App learned that Cathay Pacific Haitong Securities released a research report saying that the 2025 Volcano Engine Winter FORCE Power Conference will set up 3 main forums and 20 exclusive forums, which can focus on agent development tool upgrades and ecological expansion. The conference covered the three main forums for publishing, technology, and developers, as well as four exclusive forums for products, industries, innovation, and developers. From Al+ finance, Al+ education, AL+ automotive to HiAgent's cutting-edge exploration, from Data+Al integration to enterprise applications, etc., the AI implementation path is analyzed in a comprehensive manner.
Cathay Pacific Haitong Securities's main views are as follows:
The 2025 Volcano Engine Winter FORCE Conference is scheduled for December 18-19, focusing on the Agentic Al reshaping industry. The official announcement is that the conference will have three major releases: the Doubao Big Model Family has been completely renewed, Agent development tools have been upgraded, ecological expansion, and the ability to build a full AI implementation link. On November 25, according to a Tweet on the Volcano Engine official account, the 2025 Volcano Engine Winter FORCE Power Conference was scheduled for December 18-19 to focus on reshaping the Agentic industry, announcing that AI is moving from passive response to actively completing tasks, and the implementation value is outstanding.
The Winter Conference can continue to pay attention to the further evolution of the big bean bag model in deep thinking and multi-modal integration. The 2025 Volcano Engine Spring FORCE Conference revealed the full series of updates to Doubao Big Model 1.6, including 1.6, 1.6-thinking, and 1.6-flash models. Doubao 1.6 has improved in terms of reasoning, mathematics, instruction following, agents, etc., and the comprehensive cost is 63% lower than Doubao 1.5 Deep Thinking and DeepSeek R1. The Winter Conference can continue to focus on the further evolution of Doubao's deep thinking and multimodal integration. Combining the Spring All in One/Thinking/Flash's “hierarchical+range pricing” and low latency strategy, it will evaluate a new generation of “cost performance” optimization paths in terms of application breadth and cost.
Furthermore, according to the 2025 Spring FORCE Conference, as of the end of May 2025, the Dobao Big Model used more than 16.4 trillion tokens per day. On October 16, 2025, Volcano Engine revealed the latest token call data. The usage of the Doubao Big Model has increased 253 times from 120 billion tokens in May 2024 to over 30 trillion tokens in September this year. This Winter Force Conference may disclose the latest token call volume.
Risk warning: Large models are iterating faster than expected, the supply of computing power is insufficient, and data privacy compliance risks.