Outsourced computing is getting widespread attention in the era of cloud computing and mobile device adoption. Non-interactive verifiable computation (NIVC) with a single server has an inherent trade-off between efficiency and security/privacy, leading to an interest in multi-server NIVC. A recent study by Zhang et al. (S&P2022) introduced efficient schemes for outsourcing polynomials. However, they require a linear relationship between the number of servers and the degree of the outsourced polynomial and do not account for function privacy, thus constraining their applicability. To remove these limitations, we propose a multi-server public verifiable computation (MSPVC) model and a corresponding MSPVC scheme for outsourcing m-variable d-degree polynomials, ensuring computational input/function privacy and security. It supports public delegation and verification, including individual verification for each server. Moreover, it provides faster private verification for the client and enables private reconstruction. Detailed comparative experiments, including functionality and efficiency, demonstrate the superiority of our scheme.