At a developer conference held in Wuhan, capital of central Hubei province, Baidu co-founder, chairman and chief executive Robin Li Yanhong unveiled the multimodal Ernie 4.5 Turbo and X1 Turbo reasoning models.
According to Li, Ernie 4.5 Turbo costs about 40 per cent less than DeepSeek’s namesake V3 large language model (LLM), while X1 Turbo is being offered at a quarter of the price of the Hangzhou-based AI start-up’s R1 reasoning model.
“The essence of innovation is the lowering of cost,” Li said in his keynote speech, which stretched more than an hour. He pointed out that this would help developers focus on creating “more super interesting and useful” AI applications, rather than the expense involved in a foundational AI model.
Baidu’s latest AI models were introduced just weeks after the Beijing-based company made its Ernie Bot service free to use from April 1, providing wider access to its AI models, as competition in this nascent field heats up with the rise of innovative newcomers like DeepSeek.
