TrustedRouter launches three new encrypted 'fusion' models, including US one (venturebeat.com)

TrustedRouter has announced the launch of three new encrypted "fusion" models—Prometheus 4.0, Liberty 4.0, and Polyphemus 1.0—as part of its effort to streamline AI model selection for enterprises while ensuring data privacy. These models allow developers to route requests intelligently among various AI providers based on specific task requirements, cost, and privacy settings. Notably, the architecture maintains encrypted connections, ensuring that user prompts remain secure until they reach the processing hardware, safeguarding sensitive information from providers' operators. The significance of TrustedRouter's announcement lies in its potential to enhance the efficiency and cost-effectiveness of AI model utilization. Each model functions with distinct routing mechanisms: Prometheus and Liberty combine responses from multiple models for improved accuracy, while Polyphemus selects a single model for efficiency. By leveraging confidential computing technologies powered by NVIDIA GPUs, TrustedRouter not only allows real-time comparison of model performance but also positions itself as a flexible solution for enterprises navigating the complexities of AI workloads, potentially reducing costs by about 60% compared to prominent models like Claude Fable 5.1. This innovation could reshape how companies engage with AI technologies, optimizing for both quality and cost in a rapidly evolving marketplace.