According to the research report released by Kuaishou Securities, based on data from Volcano Engine, the average daily token usage of the Doubao large model in December exceeded 4 trillion, growing more than 33 times since its release in May. The invocation of the Doubao large model in scenarios such as information processing, customer service, sales, hardware assistants, and AI tools is also rapidly increasing. The growing invocation and multi-scenario coverage make the Doubao large model increasingly comprehensive, leading to a comprehensive upgrade. The Doubao general model pro has completed a new version iteration, with its overall task processing capability improved by 32% since May. The continuous iteration of the Doubao large model will accelerate the implementation of AI applications in multiple scenarios and expand the space for commercialization.

KYG Securities' main points are as follows:

ByteDance releases the Doubao visual understanding model, which is expected to land in multi-scenario applications.

On December 18, the Volcano Engine under ByteDance released the Doubao visual understanding model at the 2024 FORCE Power Conference. Through the Doubao visual understanding model, users can input both text and related questions about images simultaneously. The model can comprehensively understand and provide accurate answers, significantly simplifying the development process.

The Doubao visual understanding model mainly has three capabilities: (1) stronger content recognition capability, which can not only identify the categories and shapes of objects in images, but also understand the relationships between objects, spatial layout, and overall scene meaning; (2) stronger understanding and reasoning ability, which can not only recognize graphic and text information but also perform complex logical calculations; (3) more refined visual description capability, which can provide nuanced descriptions of the content presented in images based on image information and can create various writing styles.

Based on these capabilities, the Doubao visual understanding model has wide applications in scenarios such as Education, tourism, and e-commerce. For example, in the Education scenario, it optimizes essays and popular science knowledge for students; in the tourism scenario, it helps tourists read foreign menus and provides background knowledge of buildings in photographs; in the e-commerce marketing scenario, it assists merchants in fully describing Commodity details and efficiently publishing promotional ads, etc. Additionally, the input price for Doubao visual understanding is 0.003 yuan per thousand tokens, which is 85% lower than the industry average price, benefiting enterprises and developers in utilizing the visual understanding model to create commercial value in a wider range of scenarios.

The usage of the Doubao large model has increased significantly, expanding the model family, and enhancing multi-modal capabilities continuously.

According to data from Volcano Engine, the average daily tokens usage of the Doubao large model in December exceeded 4 trillion, growing more than 33 times since its release in May. The model's application in information processing, customer service and sales, hardware assistance, AI tools, and other scenarios is also rapidly increasing. The continuously rising call volume and multi-scenario coverage make the Doubao large model more comprehensive, leading to a complete upgrade, in which the Doubao general model pro has completed a new version iteration, improving its comprehensive task processing capability by 32% compared to May.

In addition to the visual understanding model, Volcano Engine has also released Doubao music model 4.0, Doubao text-to-image model 2.1, and veOmniverse + Doubao 3D generation model. The Doubao video generation model will officially open services in January 2025, and next spring, ByteDance will launch Doubao video generation model version 1.5, which will have the ability to generate longer videos. Furthermore, Volcano Engine has introduced a comprehensive AI search feature that closely integrates the enterprise's information, business, and user needs through scenario-based search recommendation integrated services, private domain information integration services, and connected question-and-answer services, accelerating the intelligent transformation across multiple industries.

The Doubao large model is expected to drive rapid development in the large model industry, and attention should be paid to related AI application investment opportunities.

The training of the Doubao multi-modal model may drive the demand for text, image, and 3D material corpus, with a strong recommendation for Funshine Culture Group. Benefiting symbols include Visual China Group (000681.SZ), Silkroad Visual Technology (300556.SZ), Tianyu Digital Technology (002354.SZ), COL Group Co.,Ltd. (300364.SZ), and IReader Technology (603533.SH); the Doubao music model may accelerate user penetration in AI music, with a strong recommendation for Hubei Century Network Technology Inc. (300494.SZ); the Doubao video generation model may expedite the production of film and television content and reduce costs for IP monetization, with a strong recommendation for Shanghai Film (601595.SH), benefiting symbols including Beijing Jetsen Technology (300182.SZ), Zhejiang Huace Film & TV (300133.SZ), and Beijing Enlight Media (300251.SZ).

The Doubao visual understanding model may accelerate the commercialization of multi-scenario AI applications: AI + E-commerce/Marketing, with a strong recommendation for MOBVISTA (01860) and benefiting symbols including Inly Media Co., Ltd (603598.SH), Guangdong Insight Brand Marketing Group (300781.SZ), and Foshan Yowant Technology (002291.SZ); AI + Companionship/Toys, with a strong recommendation for Alpha Group (002292.SZ) and benefiting symbols including Zhejiang Jinke Tom Culture Industry (300459.SZ); AI + Education Publishing, benefiting symbols include Astro-century Education & Technology (300654.SZ), Beijing Shengtong Printing (002599.SZ), and Southern Publishing and Media (601900.SH).

Risk warning: The iteration speed of the Doubao large model may be below expectations; the commercialization process of Doubao AI applications may not meet expectations.