Based on the four dimensions of innovation capacity， innovation quality， innovation effectiveness， and innovation contribution， a modernization evaluation system for higher education institutions is constructed to measure the comprehensive level of knowledge innovation within urban agglomerations， considering both local contributions and cross-city spillovers. Methods such as generalized extreme value distribution fitting and unconditional distribution estimation are employed to analyze the distribution patterns and spatiotemporal evolution characteristics of city scores. Furthermore， the Dagum Gini coefficient decomposition is used to explore the sources of disparities， and spatial econometric models are applied to examine spatiotemporal convergence and the evolutionary mechanisms driven by leading cities. The study finds that the level of urban knowledge innovation exhibits a significant right-skewed heavy-tailed distribution， with polarization trends intensifying over time. Spatial differences primarily stem from inter-regional disparities among urban agglomerations. Within urban agglomerations， conditional convergence trends and significant spatial spillover effects from leading cities are observed， indicating that the evolutionary process is jointly driven by catch-up effects and radiation effects. This study attempts to reveal the overarching characteristic of knowledge innovation in Chinese cities—"overall disparities entrenched in regional blocks， with local evolution driven by both convergence and spillover effects." It provides an empirical basis for the integrated advancement of education， technology， and talent development during the 15th Five-Year Plan period， as well as for strengthening the effectiveness and contribution of higher education institutions to regional innovation.