An illustration of using AI applications on a mobile phone. Photo: VCG

China's tech companies are ramping up efforts in artificial intelligence (AI) development, launching a growing number of new models following DeepSeek's R1 release earlier this year that gained international fame. 

Baidu launched its latest foundation models - ERNIE 4.5 and ERNIE X1 - on Sunday, highlighting the intensifying AI contest in China. ERNIE 4.5 is Baidu's latest self-developed multimodal model, and it achieves collaborative optimization through joint modeling of multiple modalities, demonstrating exceptional multimodal comprehension capabilities, according to information the company provided to the Global Times on Sunday.

As the first multimodal deep-thinking reasoning model capable of tool use, X1 also excels in Chinese knowledge questions and answers, literary creation, manuscript writing, dialogue, logical reasoning and complex calculations, according to the company.

Baidu is not alone in launching new AI models. Previously, on March 6, Alibaba Cloud released and open-sourced its new inference model, Tongyi Qianwen QwQ-32B. The Tongyi Qianwen QwQ-32B model strikes a balance between performance and efficiency, the Xinhua News Agency reported.

It makes significant strides in mathematics, coding and general capabilities, matching the overall performance of DeepSeek-R1, the report said. Additionally, while maintaining a strong performance, QwQ-32B greatly reduces deployment costs and can be locally deployed on consumer-grade graphics cards.

Another Chinese tech giant, Tencent, introduced Hunyuan Turbo S on February 27. This model provides near-instant responses, doubling text output speed and reducing initial delay by 44 percent, Tencent said in an online post.

Domestic large models, primarily trained in Chinese, capitalize on language advantages, positioning them to offer distinct strengths in areas like comprehension and creativity, Chen Jing, vice president of the Technology and Strategy Research Institute, told the Global Times on Sunday.

China's development of large AI models benefits from several key advantages compared with international markets, including strong original research capabilities and a flourishing open-source ecosystem, he said.

"Our strengths lie in our large population, abundant data resources and diverse use scenarios," Chen said, adding that Chinese companies also excel in data collection, processing and annotation, enhancing their competitive edge in practical applications.

China has introduced various policies to support and regulate AI growth, with the country's 2025 Government Work Report also emphasizing large AI model adoption.

"Under the AI Plus initiative, we will work to effectively combine digital technologies with China's manufacturing and market strengths. We will support the extensive application of large-scale AI models and vigorously develop new-generation intelligent terminals and smart manufacturing equipment, including intelligent connected new-energy vehicles, AI-enabled phones and computers, and intelligent robots," the report said.

Chinese localities like Beijing, Shanghai and Guangdong are also making efforts to build AI industry hubs.

Shanghai plans to strengthen AI development with an open and collaborative innovation framework, the 21st Century Business Herald reported. At the 2025 Global Developer Conference, Shanghai is committed to expanding AI cooperation, fostering open-source development, and promoting innovation and data-sharing, the report said.

Guangdong has rolled out 12 AI and robotics related policies to accelerate industrial digitalization, the Nanfang Daily reported last week. Under the new policy, Guangdong will fund up to 10 AI and robotics projects annually, offering subsidies of up to 8 million yuan ($1.11 million) per project.

The scale of China's AI sector is projected to reach 811 billion yuan by 2028, according to a report from iResearch.