China’s artificial intelligence sector has entered a new phase of rapid competition, with multiple companies releasing advanced models ahead of the Lunar New Year. Firms including Alibaba, ByteDance, Zhipu, MiniMax, Moonshot, and DeepSeek are pushing aggressively into global markets, while analysts warn that the same tools may also carry political influence aligned with Chinese government priorities.
The latest wave shows how quickly China is closing the gap with the United States in AI capability, cost, and adoption.
Latest Chinese AI Models and Their Capabilities
Alibaba Qwen 3.5
Alibaba’s newest model is multimodal across more than 200 languages and designed for agent automation. The company says it can deploy AI agents up to five times faster than earlier models, allowing systems to fill forms, navigate websites, and complete complex workflows. Early testing shows the model can generate functional 3D games and analyze medical images, while being up to 60 percent cheaper than its predecessor.
ByteDance Doubao 2.0
Doubao 2.0 is the latest version of ByteDance’s chatbot platform with improved reasoning and multi step task execution. Reports indicate performance comparable to leading Western systems like ChatGPT and Gemini. The model is designed to power digital assistants that can execute decisions autonomously, reflecting China’s push toward AI agents capable of handling everyday digital tasks without human supervision.
ByteDance SeeDance 2.0
SeeDance 2.0 focuses on video creation, allowing users to generate immersive audio and video with professional level controls. The tool supports both text to video and image to video generation. Critics, including the American Motion Picture Association, warned it could enable large scale copyright infringement, stating the platform operates “without meaningful safeguards against infringement.”
Zhipu GLM 5
GLM 5 is an open source model built for advanced reasoning, coding, and long task execution. It can create assistants that browse, call tools, and manage workflows across long sessions. The system uses DeepSeek’s sparse attention method to reduce computing costs and was reportedly trained entirely on Huawei chips, which the company called a milestone in “self reliant AI infrastructure.”
MiniMax M2.5
MiniMax’s M2.5 has drawn strong attention from analysts because its performance rivals leading Western models while costing far less. UBS analysts reported usage reached about one third of Anthropic’s Claude at one tenth the price. The company also offers video generation, audio synthesis similar to ElevenLabs, and AI companionship tools, positioning itself as a broad multimedia AI platform.
Moonshot Kimi 2.5
Kimi 2.5 focuses heavily on coding and agent driven task completion. The model emphasizes productivity and automation rather than entertainment or audiovisual features. It is designed to complete complex digital tasks through autonomous planning and execution, reflecting a major trend toward AI agents that act independently on behalf of users.
DeepSeek V4 (Expected)
DeepSeek’s upcoming model has generated global anticipation after its previous release shocked markets with advanced performance despite U.S. chip restrictions. The new version is expected to replace V3, which previously became the most downloaded free app globally. Recent upgrades to its context window suggest significantly improved memory and task handling.
Potential Markets and Economic Opportunity
Chinese companies are targeting both domestic and international markets with aggressive pricing and integration strategies. UBS analysts estimate MiniMax alone could capture 3 percent of the global enterprise market, representing about 41 billion dollars in revenue. Video generation could add another 5 billion dollars, while AI companionship may represent a 4 billion dollar market.
Chinese internet giants are also pushing adoption through consumer apps. Analysts said Lunar New Year promotions will “accelerate AI penetration among users especially in lower tier cities” and promote functions like image generation, quick commerce, and transactional bookings within AI platforms.
The strategy combines entertainment, productivity, and commerce into unified ecosystems designed to attract massive user bases quickly.
How China Uses AI Models to Spread Propaganda
Security researchers warn that Chinese AI systems often contain embedded political controls aligned with government narratives.
An Estonian intelligence report found that when discussing security issues, the DeepSeek model “conceals key information and inserts Chinese propaganda into its answers.” Studies also showed distortions related to Russia’s invasion of Ukraine, including unprompted references to Chinese government positions such as statements that China supports peace and dialogue.
Researchers discovered internal instructions directing models to keep content about China “positive and constructive, avoid criticism, and emphasize achievements.” These controls can spread beyond the original models into downstream applications built on them, sometimes without developers realizing the embedded guidance.
Chinese officials view AI exports as a strategic tool to expand influence globally. Scholars have discussed using AI advances to “command greater discourse power on the international stage.”
Competition With the United States
China is widely seen as the primary competitor to the United States in artificial intelligence. The rapid release cycle and lower pricing create strong competitive pressure on American companies.
Chinese open source models have already surpassed other countries in downloads on the Hugging Face platform. DeepSeek alone has been downloaded more than 75 million times. Some governments, including Italy and Denmark, have restricted the technology due to security concerns, highlighting geopolitical tensions around AI adoption.
There are also strategic industrial moves underway. DeepSeek reportedly gave early access to Chinese chip makers like Huawei while excluding U.S. firms such as Nvidia and AMD, which one analyst said may reflect a broader government strategy “to try to keep U.S. hardware and models disadvantaged in China.”
Despite progress, analysts note that U.S. firms still lead in many areas, particularly advanced semiconductor design and frontier research. However, China’s lower cost models and aggressive deployment strategies are narrowing the gap.
Analysts see both opportunity and disruption coming from China’s AI expansion. UBS said it is “positive on MiniMax’s potential upside in gaining share in the global enterprise market,” reflecting investor optimism about China’s role in AI growth.
At the same time, Western security researchers warn that the global spread of Chinese AI could reshape the information environment. Reports emphasize that democracies must increase transparency and safeguards because AI systems with embedded political bias could influence elections, public opinion, and global discourse.
The rapid pace of development suggests the competition between China and the United States is entering a new phase where cost, scale, and influence may matter as much as raw technological performance.
If current trends continue, the world may soon face not just an AI race, but a battle over whose artificial intelligence shapes the global flow of information.
But remember the downside, that the more China wins, the greater their propaganda influence.