Baidu releases the official version of its Wenxin Big Data Model 5.0: 2.4 trillion parameters, with full-modal understanding and generation capabilities.

Baidu officially released version 5.0 of its Wenxin Big Model, boasting 2.4 trillion parameters. Employing native full-modal unified modeling technology, it possesses comprehensive capabilities for understanding and generating information from text, images, audio, and video. The official statement indicates that in over 40 authoritative benchmark tests, its language and multimodal understanding capabilities surpass those of mainstream international models such as Gemini-2.5-Pro and GPT-5-High.