ChatYuan is an open-source dialogue large language model from China’s ClueAI team, built for both Chinese and English conversation in a ChatGPT style. Its large-v2 release focuses on delivering strong chat quality at a lightweight size โ small enough, after INT4 quantization, to run inference on consumer GPUs, PCs and even phones. It is an open-source project on GitHub with around 1,900 stars; the models were released in 2023.
ChatYuan emerged from the Clue benchmark community, one of China’s earliest open NLP evaluation efforts. The v2 model refines the original with better fine-tuning data, reinforcement learning from human feedback and chain-of-thought behaviour, targeting practical bilingual dialogue rather than benchmark chasing. Weights are distributed through Hugging Face and ModelScope with online demos, and the project documents local deployment paths for modest hardware, reflecting its goal of accessible Chinese-language conversational AI in the first wave of open chat models.
Free. The ChatYuan models are released openly for download and local use; check the model card for the applicable usage terms.
ChatYuan sits in the ClueAI model family alongside PromptCLUE and the Clue benchmark suite, and its releases were notable in early 2023 as some of the first openly available Chinese chat models following ChatGPT’s debut. The large-v2 checkpoint improved on the first release’s dialogue consistency and instruction following, and the project pages document prompt formats for getting the best behaviour from the model in both Chinese and English conversations.
Developers and researchers interested in compact bilingual chat models, Chinese NLP history, or running a small self-hosted Chinese assistant on limited hardware.