Release of the 'Fangsheng' Large Model Benchmark Results (Q1 2025) and Overview of Baidu's Wenxin 4.5 and X1 Models

The China AI Industry Alliance (AIIA) has closely tracked the development of large models and intelligent agents. It built and released the "Fangsheng" (FactTesting) benchmark system, conducting six rounds of capability monitoring since 2024 and testing over 200 open‑source and proprietary models. In 2025 the scope was expanded to multimodal understanding, text‑to‑image, text‑to‑video, and early research on intelligent‑agent testing, establishing a verification platform for the industry.

On April 9, 2025, at the 14th plenary meeting in Nanjing, AIIA officially announced the Q1 2025 "Fangsheng" benchmark results. In the large‑language‑model results, Baidu's Wenxin 4.5 ranked highest in basic abilities, while Wenxin X1 led in reasoning abilities.

On March 16, 2025, Baidu released Wenxin 4.5 and Wenxin X1. Wenxin 4.5 is a new generation native multimodal foundation model that integrates multiple modalities for collaborative optimization, offering superior multimodal understanding, enhanced language capabilities, reduced hallucinations, stronger logical reasoning, and notable code generation ability.

Wenxin X1 possesses stronger comprehension, planning, reflection, and evolution capabilities, supports multimodal input, and is the first deep‑thinking model that autonomously utilizes tools. It demonstrates outstanding performance in Chinese knowledge Q&A, literary creation, document writing, everyday dialogue, logical reasoning, complex calculations, and tool invocation.

Both models are now available for free on the Wenxin Yiyan website (https://yiyan.baidu.com). Baidu plans to make bold investments in artificial intelligence, data centers, and cloud infrastructure throughout 2025 to build the next generation of smarter models.