Baidu has introduced its latest foundation AI models, ERNIE 4.5 and ERNIE X1, making them freely available for individual users through ERNIE Bot. This move aims to push the boundaries of multimodal reasoning and AI models while ensuring greater accessibility at a lower cost. The company plans to integrate these advanced models into its product ecosystem, including Baidu Search and the Wenxiaoyan app, to enhance user experiences.
As Baidu’s newest multimodal foundation model, ERNIE 4.5 leverages collaborative optimization across different modalities, improving comprehension in text, images, audio, and video. It enhances language understanding, content generation, reasoning, and memory while significantly reducing AI hallucinations. The model demonstrates strong contextual awareness, enabling it to process complex media like internet memes and satirical cartoons. Compared to GPT-4.5, ERNIE 4.5 reportedly delivers superior performance at just 1% of the cost. These advancements stem from innovative technologies such as FlashMask dynamic attention masking, multimodal mixture-of-experts, spatiotemporal representation compression, and knowledge-centric training methodologies.
Alongside ERNIE 4.5, Baidu has introduced ERNIE X1, a deep-thinking reasoning model designed for advanced problem-solving, strategic planning, and AI-driven evolution. As Baidu’s first multimodal model capable of tool use, ERNIE X1 excels in areas such as Chinese knowledge-based Q&A, literary creation, and complex mathematical calculations. It incorporates advanced search capabilities, document Q&A, image recognition, AI-generated imagery, and webpage reading functions. The model benefits from progressive reinforcement learning, an end-to-end training approach integrating chains of thought and action, and a unified multi-faceted reward system.
For businesses and developers, ERNIE 4.5 is now accessible via APIs on Baidu AI Cloud’s Qianfan platform, with competitive pricing. ERNIE X1 will soon be available on the same platform. Baidu anticipates that 2025 will be a pivotal year for AI evolution and plans to continue investing in AI research, data centers, and cloud infrastructure to drive the next generation of AI innovation.
