Last released Sep 9, 2026
Low-downtime hot migration for vLLM inference services, via monkey-patching and a plugin (no vLLM source changes).