Domestic AI chip software stacks accelerate support for open models

Summary

Inference frameworks added operator optimizations aimed at reducing enterprise migration costs for private deployments.