Baidu, a Chinese company, announced on Sunday the launch of two new artificial intelligence models in Hong Kong. One of the models focuses on reasoning and is being positioned as a competitor to DeepSeek’s model in the highly competitive AI landscape. DeepSeek, another Chinese AI startup, has caused a stir in the industry by offering AI models that rival or surpass those of leading U.S. companies at a lower cost, fueling the global AI race.
Baidu’s new model, ERNIE X1, is said to match the performance of DeepSeek R1 but at half the price. It boasts enhanced capabilities in understanding, planning, reflection, and evolution, being the first deep thinking model to autonomously utilize tools. Additionally, Baidu introduced its latest foundation model, ERNIE 4.5, which excels in multimodal understanding and features advanced language skills along with improved logic, memory, and generation abilities.
Described as having a “high EQ,” ERNIE 4.5 is adept at comprehending network memes and satirical cartoons. Despite Baidu being one of the first tech giants in China to launch a ChatGPT-style chatbot, its Ernie large language model has faced challenges in gaining widespread adoption, despite claiming performance comparable to OpenAI’s GPT-4.
Multimodal AI systems, which can process and integrate various data types like text, video, images, and audio, play a crucial role in content conversion across different formats. Baidu’s developments in this area underline its commitment to staying competitive in the rapidly evolving AI landscape.
