DeepSeek, a Chinese artificial intelligence startup, has launched an updated version of its foundation model, V3.1-Terminus. The update focuses on enhancing the model’s agentic capabilities, allowing it to interact with simulated environments and perform tasks more effectively. The company acknowledges incorporating user feedback to improve the model’s performance. V3.1, previously recognized as the most advanced model by Artificial Analysis, has been upgraded with improvements in navigation and decision-making. DeepSeek plans to continue refining V3.1 through ongoing testing and development, focusing on expanding the model’s operational scope.
Credits: News – South China Morning Post