Chinese tech giant Baidu took a big step on Wednesday to redefine its core services, launching new AI video generator, MuseSteamer, along with the most important overhaul of search engines in a decade. The twin release, announced by Beijing, leverages Baidu's powerful Ernie AI to transform the platform.
The goal is to move from a simple information tool to an interactive, task-oriented assistant. This strategic push is aimed at fighting fierce competition with domestic rivals such as Bytedance and global players such as Google. It is also based on open source AI from Baidu's recent offensive pivot.
This demonstrates the clear intention to compete in both the capabilities and costs of an escalating global AI race.
From search boxes to smart assistants
Baidu's search platform has undergone the most drastic change in years. The company expanded its mobile app search bar to replace it with a “smart box” equipped with the Ernie AI model. This new interface supports long text input of thousands of words. This is a big leap from previous limitations.
Upgrades expand searches beyond simple text queries. It currently handles audio and image-based searches and integrates over 18,000 third-party AI agents. This allows the platform to perform complex tasks such as planning trips, generating AI images, and even creating research reports.
This evolution reflects Baidu's ambition to expand the “search boundary” and moves from a tool that retrieves information to one that helps users complete tasks directly within the app. The changes will be rolled out in stages over the coming weeks.
MuseSteamer enters AI Video Arena
At the same time, Baidu announced MuseSteamer. This is an image-to-video model aimed at business users. This tool can generate video clips up to 10 seconds in length and is available in three versions: Turbo, Pro and Lite. Unlike its consumer rivals, Baidu has not yet released any public apps.
To promote the new tool, Baidu launched a creative contest on the Huixiang platform, encouraging users to generate “pinch and play” videos, a popular meme style in China. The move pushes Baidu into crowded areas against global technology Titans such as Google, Microsoft and Meta.
All companies have recently launched or upgraded their own video generation tools, such as Google's VEO 3 and Bing Video Creator powered by Microsoft's free SORA. The competition is fierce, with each player competing for dominance in a market that is expected to grow significantly.
A calculus war on price and openness
The launch of these products is the latest salvo in Baidu's aggressive new strategy. Just a few days ago, the company announced that it was creating Ernie AI Model Family Open Source.
This pivot is a direct challenge to a high-cost model of Western leaders like Openai. Analyst noted, “Baidu just throws Molotov into the AI world. This is not a competition. It's a declaration of war on pricing.” The pricing war began earlier this year, when Baidu made Arnie Bot free and reduced the API costs of the model by more than 80%.
Baidu CEO Robin Li explained this strategy as a way to empower developers. “Why are they continuing to lower the prices of large language models? They can't afford it because it's a big obstacle for developers to write AI applications because of high costs.” He said at the company's developer meeting. This embrace of accessibility shows a major shift from his previous skepticism about the open source movement.
Geopolitical chess and trust deficits
Baidu's open source gambit is also a powerful tool in the escalation of the US-China tech war and could potentially avoid US sanctions. However, this strategy is seen in deep doubt in Washington. Washington sees China's open source AI as a national security risk.
American officials issued harsh warnings. Recently, the House Selection Committee on CCP, known as rival Chinese AI company Deepseek “A weapon of the Communist Chinese weapons designed to spy on Americans, steal our technology and overturn US laws.”. This highlights concerns about data privacy and the potential misuse of technology.
Experts warn that transparency remains a major issue. As USC Professor Shaun Wren warns, “Just because the model weights are publicly available doesn't mean that we know which data was trained, whether consent was given, or whether those data contributors were credited or compensated.”. This trust deficit can hamper Baidu's ambitions as it competes not only for features and prices but also for security and responsible development promises.
