DeepSeek open-sources AI inference optimizations, 60-85% faster generation

This development could accelerate the adoption and reduce the operational costs of generative AI models, potentially impacting cloud providers and enterprises deploying AI at scale by making AI inference more efficient and accessible.