Dragonfly 2.5.0 enables P2P model pulls and auto‑injects into Kubernetes pods
Dragonfly 2.5.0 adds direct AI model repo downloads from Hugging Face and ModelScope using Git‑LFS‑accelerated P2P, and ships a Kubernetes mutating webhook that injects the client into pods automatically. Teams can now fetch large models without rebuilding images, cutting latency and operational overhead for ML workloads.