ModelCascade – Route LLM calls to your own GPU first, cloud second

(github.com) by wayneIA | Apr 15, 2026 | 1 comments on HN