> For the complete documentation index, see [llms.txt](https://brindha.gitbook.io/mylearning/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://brindha.gitbook.io/mylearning/tools/baichuan.md).

# Baichuan

**What it is**

Baichuan Intelligence is a Chinese AI company founded by Wang Xiaochuan on April 10, 2023, focusing on the research and development of general AI foundational large models and their upper-layer applications. Its core team is composed of top AI talents from companies such as Sogou, Baidu, Huawei, Microsoft, ByteDance, and Tencent. 

**Model lineup**

Baichuan Intelligence has released several large models including Baichuan-7B, Baichuan-13B, Baichuan-53B, Baichuan 2-53B, Baichuan 2-192K, Baichuan 3, and Baichuan 4, which have excellent performance in Chinese natural language processing, mathematics, and coding capabilities.

**Training scale**

Baichuan 2 was trained on a high-quality corpus with 2.6 trillion tokens and achieved the best performance of its size on multiple authoritative Chinese, English, and multi-language general and domain-specific benchmarks.

**Chinese language leadership**

Baichuan 4 is ranked highest on SuperCLUE, a benchmark test that specializes in evaluating Chinese language models, placing it above GPT-4 Turbo and Claude 3 Opus in measuring Chinese language capabilities.

**Medical specialization**

Baichuan-M3 is a medical-enhanced large language model trained to explicitly model the clinical decision-making process, proactively acquiring critical clinical information, constructing coherent medical reasoning pathways, and systematically constraining hallucination-prone behaviors. It surpasses GPT-5.2 across HealthBench, HealthBench-Hard, hallucination evaluation, and clinical benchmarks.

**Funding**

In July 2024, the company completed a Series A financing round of 5 billion yuan, reaching a valuation of 20 billion yuan.