Baidu, a Chinese search engine company, announced on Sunday that it has released two new artificial intelligence models, including one focused on logical thinking that can compete with the model from DeepSeek in the highly competitive AI market.
Both models are free for personal use and the company plans to make its AI platform available as open source to other developers and companies. According to Baidu, the new model, ERNIE X1, offers the same performance as DeepSeek R1 at half the price and has stronger understanding, planning, reflection and evolution capabilities. The company also claims that it is the first deep-thinking model to autonomously use tools.
Baidu’s new basic model, ERNIE 4.5, is said to have excellent multimodal understanding capabilities and is expected to have improved language abilities, with enhanced understanding, generation, logic and memory capabilities. The model is also described as having a high emotional quotient, making it easy to understand in the context of internet memes and satirical cartoons.
As one of the first Chinese tech giants to release a chatbot in the style of ChatGPT, Baidu has faced challenges in promoting its large language model, Ernie, despite claims of its comparable performance to OpenAI’s GPT-4.
Multimodal AI systems, like Baidu’s ERNIE, are capable of processing and integrating various types of data, including text, video, images and audio and can convert content between these formats.
The competition in the AI technology has recently reached a new level, following the success of a Chinese AI model in late January, which triggered a market shock. The model, DeepSeek, was able to match the technology of American companies like OpenAI at a fraction of the development costs, leading to a loss of over a billion US dollars in the value of US and European technology stocks. The US government has already cautioned American companies not to fall behind in the AI race.