On June 17th, the 2026 Meitu Imaging Festival was held in Xiamen, where four new products—Picchi, Artflo, MVLAND, and MeituHub—were launched, alongside upgrades to existing platforms like Zcool, Meitu Design Studio, Kaipai, and RoboNeo.
Examined within the context of Meitu's AI evolution over the past two years, this latest shift places a stronger emphasis on delivering tangible results.
While the company previously focused on providing features for photo editing, shooting, and design, this year's product lineup distinctly moves towards offering complete outcomes. Services like portrait retouching, talking-head videos, marketing materials, music visuals, and AI-generated short dramas are now packaged as ready-to-use, deliverable workflows.
CEO Wu Xinhong noted in a group interview that user needs are evolving in the AI era. With a proliferation of AI applications, many users are not inclined to invest time in learning complex systems. Reflecting this, the latest products are designed so that users don't necessarily need to master prompts, parameters, or intricate workflows; instead, paid usage is increasingly directed towards immediately usable images, videos, and commercial assets.
Transitioning from Agents to Agent Teams: A Focus on Outcomes Over Features
Picchi targets advanced portrait retouching, MVLAND serves musicians and labels, Kaipai focuses on talking-head and marketing videos, Meitu Design Studio is evolving into an AI design team, and RoboNeo is experimenting with AI short dramas. These products, though serving different scenarios, share a unified commercial direction: minimizing the addition of functional buttons for users and maximizing the delivery of specific, finished results directly into their hands.
A recent Goldman Sachs report on Meitu summarized its transformation as an upgrade from a "beautification tool" to a "professional content engine." The report suggests generative AI is propelling the company from the consumer entertainment market into the enterprise productivity tools market, projecting a 29% compound annual growth rate for revenue from 2025 to 2030, with the share of productivity tool revenue rising from 12% in 2025 to 44% by 2030.
At the launch event, more direct signals regarding commercialization came from CFO Yan Jinliang, who highlighted two key data points.
Within Meitu Design Studio, some users are already spending over ten thousand yuan monthly on AI computing power credits. Meanwhile, MVLAND has recently seen a monthly ARPU of three to four hundred yuan, approximately 20 times that of Meitu Xiuxiu.
For a company whose core business has long been C-end imaging tools, this data is more revealing than product specifications. It indicates that AI productivity products are unlocking new, higher tiers of user spending.
Meitu is packaging specific scenarios into dedicated products for user delivery: Kaipai for talking-head and marketing videos, Picchi for users requiring deep editing, MVLAND for music visuals, and RoboNeo for short drama creators.
Explaining this approach, Wu Xinhong stated, "Each of our products has its own positioning; we aim to perfect a specific scenario." He further emphasized that Kaipai is determined to dominate the talking-head video segment globally.
This product segmentation aligns with the commercial logic of AI applications.
While foundational models provide the underlying generative capability, the付费 value in imaging products often emerges within specific scenarios. Users pay not only for model calls but also for aesthetic standards, templates, assets, workflows, industry understanding, and the final outcome quality.
A J.P. Morgan report on China's AI application landscape suggests that in some enterprise contexts, access to base models may become more interchangeable, with customer value increasingly dependent on task completion, workflow integration, proprietary data, and deployment quality.
For Meitu, model capability is merely the entry ticket. Whether a product can integrate into a user's workflow determines the ceiling for user retention and付费.
On the technical front, the company's "Agent Teams" solution enables分工协作 and multi-stage verification among multiple single agents, capable of handling complex workflows to achieve better商业 outcome delivery.
Meitu's vertical products also draw需求 from within its mature offerings. Kaipai originated from the teleprompter feature in Meiyan Camera, while Meitu Design Studio grew out of the poster design function in Meitu Xiuxiu.
In a group interview, CPO Chen Jianyi noted that the commonality of these two products lies in their clear demand and the team's deeper understanding of users. Many product and operations members within the Kaipai team are content creators themselves, while Meitu Design Studio identifies typical pain points through offline exhibitions and user research.
Integrating Models into the Business Loop
Addressing discussions around "models吞噬 applications," CFO Yan Jinliang offered a response during the group interview.
He believes this assertion does not hold in the visual domain because subjective aesthetic preferences require products and services to perform final-stage calibration. While a model can generate an image, whether a user finds it natural, premium, suitable for发布, or effective for sales still depends on aesthetic standards, product refinement, and real user feedback.
Meitu has adjusted its organization accordingly. Xu Jun, head of the Meitu Design Center, explained that the company has a design center with over 200 personnel. Designers first create samples for the AI, teaching it knowledge about aesthetically pleasing AI images and AI short dramas, and then feed user feedback back into AI效果 training. The content delivered by designers is also evolving, extending from images and videos to models, creative calls, and workflows.
Model selection is also beginning to服从 business validation. Data disclosed at the event showed that from January to May 2026, an average of 96.3% of generative AI function calls within Meitu's imaging products originated from the company's self-developed Meitu奇想大模型.
In an interview, Liu Luoqi, head of the Meitu Imaging Research Institute (MT Lab), stated that the self-developed large model will be more closely integrated with the company's business products to serve deeply customized scenarios. For relatively fixed scenarios, third-party large models might be utilized.
Chen Jianyi provided a more practical formula: "the研发 depth of the model multiplied by the商业 value of the corresponding product." When a scenario's付费 potential is unproven, external models can be used for experimentation. Once商业 value is demonstrated, investment in self-developed model capabilities can follow. This sequence of investment aligns more closely with the operational logic of an application company than a单纯 emphasis on self-research.
The impact of AI on Meitu's business will ultimately manifest in its商业模式. In the past, the company primarily charged for tool functionalities like editing, filters, templates, and video enhancement. Now, productivity scenarios are beginning to drive revenue from subscriptions, AI computing power credit consumption,真人创作 services, and outcome delivery tailored to industry needs.
A recent UBS report mentioned that AI will increase the frequency of interaction between software vendors and their clients while lowering delivery costs, supporting an acceleration in software demand and margin improvement over the next 3-5 years. The report notes that traditional software companies need to choose between enhancing existing products with AI and rebuilding AI-native products from the ground up.
Viewed along this line, initiatives like the 100 million yuan product challenge also carry business significance. While Meitu boasts 280 million monthly active users, a mature imaging product matrix, and access to Zcool's 18 million designer resources, the imaging赛道 remains fragmented, and new AI-native products will continue to emerge.
Incorporating external innovation into the product funnel is, to some extent, a way to supplement its imaging ecosystem with更多垂直场景.
Judging from this Imaging Festival, Meitu's AI transformation has progressed from "adding AI to products" to "reconstructing products, organization, and收费 models with AI." Moving forward, the market will focus more on whether these applications can consistently deliver付费 imaging results. If products like Kaipai, Meitu Design Studio, and MVLAND can continue to demonstrate the replicability of their high ARPU, Meitu's valuation anchor may gradually shift from being an imaging tool provider towards becoming an imaging productivity platform.
Comments