Dracarys, a new open-source large language model (LLM) for coding, has been announced as a significant advancement in the field of generative AI. Developed to address the shortcomings of existing open-source LLMs in coding, Dracarys aims to enhance performance without compromising the model's overall capabilities. The model, classified as 70B Class, has already achieved top rankings on LiveCodeBench, an independent benchmark for test generation, where it has been recognized as the leading open-source solution for coding tasks. Released yesterday, the ability to run Dracarys locally offers organizations a high-performance option without the need to upload sensitive code to external platforms like OpenAI, Microsoft, or Anthropic.