WHAT IT IS
Baichuan is an open-weight model series Baichuan AI has released since April 2023. The company was founded in Beijing in 2023 by Wang Xiaochuan. Baichuan 2 ships in sizes such as 7B and 13B, pretrained mainly bilingually for Chinese and English and aimed at Chinese question answering, writing and knowledge tasks. Base models and chat versions are both released, with weights offered under an open licence. The series targets Chinese long-tail scenarios and is widely used by domestic developers for secondary training and local deployment.
Why it matters
It centred its open-weight models on Chinese long-tail scenarios and used small-to-mid sizes to lower the barrier to local deployment, making it a common Chinese base in domestic developers’ hands.
Key specs
- Parameters
- 7B / 13B (Baichuan 2)
- Modality
- Text
- Open weights
- Yes
- Released
- 2023-04
Capabilities
Related concepts
Transformer Architecture
Replacing word-by-word relay with a room where everyone speaks at once, so long-range dependencies are one hop away
Pretraining & Fine-tuning
Learn language first from vast unlabelled text, then specialise with little data — the most data-efficient paradigm in modern AI
Tokenization
Models do not read characters, they read tokens — and how you split text quietly sets both capability and cost
Comparable products
Qwen
2023An open-weight family spanning many sizes, with multimodal versions
Yi
2023A bilingual Chinese–English open-weight model with very-long-context versions
GLM
2023A Chinese general model that began with autoregressive blank-infilling pretraining
DeepSeek-V3
2024An open-weight MoE with 671B parameters, activating 37B per token
ERNIE
2019A Chinese model that began with knowledge-enhanced pretraining, an early landmark version