Baidu Launches New AI Model to Compete with DeepSeek

On April 25, Baidu - China's online search giant - launched two new artificial intelligence (AI) models that provide advanced multimodal inference capabilities and are priced lower than DeepSeek's equivalent products.

At a developer conference held in Wuhan, capital of Hubei province, Robin Li Yanhong - co-founder, chairman and CEO of Baidu - introduced two multi-modal inference models, Ernie 4.5 Turbo and X1 Turbo.

Among them, Ernie 4.5 Turbo is about 40% cheaper than DeepSeek's large language model (LLM) V3, while X1 Turbo is only 1/4 the price of DeepSeek's R1 inference model.

According to Li Yanhong, the core of innovation is cost reduction. This view reflects his previous prediction of exponential growth of AI applications in China this year.

Baidu's latest AI models come just weeks after the company opened up its Ernie Bot service to free use on April 1, providing broader access to its AI models as competition in the field heats up with the rise of rivals like DeepSeek.

Previously, in February, Baidu also announced that users of Ernie Bot desktop and laptop versions will have free access to Ernie LLM thanks to continuous upgrades and reduced inference costs.

LLM is the underlying technology for generative AI services like ChatGPT and Ernie Bot. In China, the mobile version of Ernie Bot is marketed under the name Wenxiaoyan.

Baidu became the first major Chinese tech company to launch an LLM in the country in March 2023, following the ChatGPT craze created by OpenAI.

However, Baidu's first-mover advantage was quickly challenged by domestic AI rivals, forcing the company to step up its efforts to enter the AI ​​field as its flagship search service faces declining traffic./.

(Vietnam News Agency/Vietnam+)

Source: https://www.vietnamplus.vn/baidu-ra-mat-mo-hinh-ai-moi-canh-tranh-voi-deepseek-post1035236.vnp