KServe

Retired

No longer on the radar

We removed this entry in the September 2026 release. Retiring is not a recommendation against it — it means the radar no longer needs a separate opinion here. The write-up below is kept as it was last published and is no longer maintained.

This decision is now covered by:

Overview

KServe standardizes Kubernetes model inference with canary rollouts, autoscaling, and multi-framework runtimes (KServe).

Assess for teams already on Kubernetes who need portable serving CRDs. Compare with managed endpoints and vLLM deployments for LLM-specific throughput.

Sources