Baidu, China’s equivalent to Google, has unveiled two new open-source AI models under the name “Ernie.” The company released Ernie X1, a reasoning model that competes with DeepSeek R1 in performance but at half the cost. Additionally, Baidu launched Ernie 4.5, a multimodal foundation model that surpasses GPT-4.5 in benchmarks while being priced at just 1% of the latter.
In a move towards open-source development, Baidu is making its chatbot, Ernie Bot, available to the public for free starting April 1, ahead of schedule. The tech giant plans to integrate Ernie 4.5 and X1 into its product ecosystem, including Baidu Search.
Baidu’s new releases come amidst discussions in Silicon Valley about the cost of AI models, particularly in response to the offerings from Chinese startup DeepSeek. Baidu’s Ernie models offer significant cost savings compared to competitors like OpenAI and DeepSeek.
The introduction of Ernie models signifies the intensifying competition between the US and China in the AI field, as well as China’s increasing adoption of open-source models to drive broader adoption. Baidu’s CEO, Robin Li, emphasized the importance of open-sourcing top models to facilitate wider adoption and drive innovation.
With China aiming to lead the global AI industry by 2030, the release of AI models like Manus and QwQ-32B from other Chinese tech companies further highlights the country’s advancements in this sector. Amidst anticipation for DeepSeek’s R2 launch, Baidu’s Ernie models are poised to challenge the status quo in the AI landscape.
