China's AIsphere unveils world's first real‑time world model for video generation — the technology has already attracted over 150 million users in 177 countries
A tech company based in Beijing's Yizhuang economic zone has become the world's first to release a universal product — an AI‑powered world model capable of real‑time video generation, which has gained recognition from users abroad. When an ordinary photo turns into a full‑fledged video in seconds, and an e‑commerce ad is generated without a film crew and post‑production, the line between reality and the digital world becomes almost invisible. Chinese company AIsphere, headquartered in Beijing's Yizhuang, has launched PixVerse R1 — the world's first universal real‑time world model for video generation.
A world model is a type of AI neural network that does not merely process text or images, but simulates physical reality and cause‑and‑effect relationships, generating videos that account for the laws of physics and spatial logic in real time. Unlike traditional models, it takes in multidimensional data (text, photos, video, motion) and allows users to dynamically influence the generated content.
This is not just another AI service — it is a new way of interacting with content, where users can influence the video feed in real time, and commercial scenarios — from advertising to the film industry — gain a tool that cuts costs and production time dramatically.
The large model developed by AIsphere has an extensive library of portrait templates, allowing ordinary users to quickly create creative scenes and characters using just one real‑life photo. In addition to photo processing, the model's most outstanding advantage is its video‑generation capability. "This is a universal world model for real‑time generation, in addition to traditional text‑to‑video and image‑to‑video models. At our company, we call it the R‑series, or Real Time. This is the world's first released universal real‑time world model," said Jin Hanmin, the company's vice president. World models can use input data such as text, images, video and motion to generate videos that simulate a realistic physical environment. Even with creative instructions detached from real‑world scenarios, the model can quickly understand and generate the desired content in real time.
The powerful AI‑based model enables the creation of videos that go beyond mere entertainment visuals and integrate deeply into practical commercial scenarios. For example, after uploading bag images, text descriptions and setting parameters such as voiceover style, the model can automatically generate a full e‑commerce promotional video. The company's product, built on the model, significantly reduces the time and costs of traditional ad shooting. As a result, many of its products have been commercialised and entered overseas markets. "On the first day of product launch, we made it available in 170 countries. Now we have more than 150 million registered users worldwide," said Jin. Jin also noted that the company's cumulative revenue in the first half of 2026 grew by 119.3% year‑on‑year, while the total number of users increased by 97.7% year‑on‑year.
Since its founding in 2023, AIsphere, led by a former head of visual technology at ByteDance, has grown from a start‑up to a global player valued at over US$2 billion. PixVerse R1, capable of generating 1080p video with instant response, has become not just a technological novelty but a platform that has changed the rules of the game in video content. Unlike traditional models that require lengthy processing, the model works in real time, allowing users to influence the narrative dynamically. Commercial success is backed by the numbers: 150 million users in 177 countries, a Series C funding round of 29.8 billion yuan (about US$440 million) with participation from Alibaba, and 119% revenue growth. The product is already widely used in e‑commerce, advertising and entertainment, and the company plans further global expansion.
As CCTV+ reports, AIsphere and its PixVerse R1 have become an example of how Chinese AI technologies are not catching up but setting trends, turning complex computational models into accessible tools for millions of users worldwide.
