Abstract: [Purpose/Significance] In the era of big science, the evaluation paradigm for science and technology innovation driven by large models is undergoing a profound change. Clarifying a number of issues on large model-driven evaluation for science and technology innovation can enhance the practical value and strategic guidance of such evaluations and provide powerful support for scientific and technological advancement and social development. [Design/Methodology] Based on the evolution of the evaluation of science and technology innovation and the main features of large model-driven evaluation for science and technology innovation, this paper deeply analyzes the main dimensions of applying large models to evaluate science and technology innovation. Furthermore, it reflects on several challenges faced by large model-driven evaluations and explores future development trends. [Findings/Conclusion] The evaluation of science and technology innovation driven by large models will become a key force for the transformation and progress of the new era's science and technology evaluation paradigm. Large models make science and technology innovation evaluation more intelligent, refined and personalized. The integrated development of large models with scientific research, technology development, science and technology evaluation, science and technology management will become an inevitable trend. [Originality/Value] A number of issues on the application of large models to science and technology innovation evaluation are analyzed to provide a reference and basis for the construction of an evaluation system for science and technology innovation with intelligence, refinement, systematization and personalization as the core and sustainability and responsibility as the orientation.

Keywords: Science and technology innovation evaluation, Large models, Digital Intelligence-drive