A phone that smooths a pimple and whitens a smile used to be the whole story. For a generation of Chinese internet users, that was Meitu: the app that made the selfie look like a magazine cover.
That business is still there. But the company that built it has quietly become something else — a workshop where people who earn a living from images and video now pay every month to get their work done faster.
The shift is not a new filter. It is a move from selling beauty to selling throughput, and the numbers behind it are bigger than the selfie ever was.
The earnings that surprised the room
When Meitu (1357.HK) reported its first-half 2026 results on 26 August, the headline was not a filter effect. Total revenue reached RMB 2.21 billion (≈ US$311M / HK$2.43bn), up 22.1% from a year earlier, while adjusted net profit hit RMB 650 million (≈ US$92M / HK$715M), up 39.5%.
Paying subscribers passed 18.44 million, a record, and the company's overall monthly active users held at about 282 million. Those figures come from the company's own interim disclosure, and they were echoed across established financial outlets including China News Service, Securities Times, and Global Times' finance desk.
The detail that matters sits deeper in the report. Meitu's image-and-design products brought in RMB 1.77 billion (≈ US$249M / HK$1.95bn), about 80% of the total, up 30.9% year on year.
From "fix my photo" to "run my workflow"
Meitu's older apps — MeituPic and BeautyPlus-style camera tools — still carry the consumer brand. But the growth engine is now a set of productivity tools aimed at people who create for a living.
The company reports that its productivity-app monthly active users reached 33 million, up 43.5% year on year, and that annual recurring revenue from those apps reached about RMB 620 million (≈ US$87M / HK$682M), up 47.8%. Tools like Kaipai (开拍), built for spoken-word and marketing video, saw their recurring revenue more than double; Vmake roughly tripled; and MVLAND, a music-video maker launched only months earlier, doubled its recurring revenue in its first quarter.
All of these are company-reported metrics, and they describe direction more than audited precision. But the pattern is consistent across Meitu's own filings and the media that covered them: the paying crowd is shifting from casual users to working creators.
The model underneath the apps
None of this works without the model layer. At its June imaging festival, Meitu released MiracleVision V6 (美图奇想大模型 V6), a mixture-of-experts large model (大模型) that takes text, image, video, and audio as inputs. The company says that between January and May 2026, about 96.3% of the generative AI (生成式人工智能) function calls inside its photo and video products ran on its own model.
That last number is a vendor claim and should be read as directional, not as independent proof. But it signals something real: the model is not a lab showcase, it is wired into the product surface where users actually click.
Charging for the result, not the tool
The quiet change in Meitu's business is how it bills. Subscriptions still lock in steady access. On top of that, the company introduced "compute-credit" consumption — users pay for the AI work they actually run, not just the right to open the app.
Meitu says credit consumption rose more than 46% quarter on quarter in both the first and second quarters of 2026. The logic is simple: a subscription prices the habit, while credits price the intensity. A creator who batches dozens of clips a day spends far more than a weekend user, and Meitu captures that gap.
It is a hedge against the old subscription trap, where light users consume full benefits while heavy users burn the platform's compute for a flat fee.
Why this matters outside China
Meitu's global monthly users have stayed above 100 million outside the mainland, and the company says overseas paid-subscriber growth is running about twice the domestic rate, led by East Asia, Latin America, and Europe. That matters because it shows a Chinese app company winning paying creators abroad on product fit, not on price alone.
For microdrama (短剧) teams, e-commerce merchants, and solo marketers, the appeal is a finished asset — a product image set, a promo clip, a music video — rather than another editing window to learn. Meitu is betting that "deliver the result" beats "here is a tool."
Honest limitations
Every revenue, user, and recurring-revenue figure above comes from Meitu's own interim results and associated voluntary disclosures, then reported by Chinese financial media; none are audited quarter-by-quarter, so they describe the company's narrative. The 96.3% self-developed-model share and the credit-consumption growth rates are vendor-reported and have not been independently verified. The product names and feature claims reflect Meitu's June 2026 festival announcements, and the overseas growth rates are the company's own framing. This article is current as of 22 October 2026 and is not investment advice.
What readers can do now
- If you create content for a living, test one result-delivery tool (Meitu's Kaipai or a Western equivalent) on a real brief and compare total time-to-finished-asset, not just output quality.
- If you run a creator tool, study the subscription-plus-credits model: it captures heavy users without punishing light ones.
- If you watch China's AI economy, track whether Meitu's overseas paid growth keeps outpacing the domestic rate through the next reporting cycle.