Hugging Face unveils Carbon-X, a conversational AI model for DNA analysis

Hugging Face announced Carbon-X, an open-source AI system that merges DNA sequence data with natural-language processing, enabling conversational queries about genetic effects. The model was co-developed with Zhongguancun Academy in China and the University of Naples Federico II in Italy, and was trained on 300 billion tokens that mix mammalian, plant, fungal and protozoan DNA with English text. With 14 billion parameters, the tool runs locally on standard laptops, preventing the upload of full human genomes and keeping user data private.

Researchers hope it will streamline the assessment of whether mutations are benign or disease-causing, supporting drug and gene-therapy design. The team deliberately excluded viral sequences to limit misuse and limited query size to roughly 50,000 base pairs. While promising, the model’s accuracy remains untested in laboratory settings.