Researchers at Hugging Face have launched a project to create an open-source version of DeepSeek’s R1 AI reasoning model. As reported by Kyle Wiggers for TechCrunch, the initiative called Open-R1 aims to duplicate all components of the original model, including training data and methods. Led by Hugging Face’s head of research Leandro von Werra, the team will use their Science Cluster with 768 Nvidia H100 GPUs for the replication. The project has already gained significant attention, receiving 10,000 GitHub stars within three days. According to Hugging Face engineer Elie Bakouch, while R1 is technically “open” with a permissive license, its development tools remain proprietary. The team expects to complete the replication within weeks, enabling researchers to build upon the technology for future open-source reasoning models.