diff --git a/k8s/infra/monitoring/servicemonitors/api-gateway.yaml b/k8s/infra/monitoring/servicemonitors/api-gateway.yaml
index bf56bd5..1a7bbfa 100644
--- a/k8s/infra/monitoring/servicemonitors/api-gateway.yaml
+++ b/k8s/infra/monitoring/servicemonitors/api-gateway.yaml
@@ -1,3 +1,25 @@
+# The gateway is the only source of per-request LLM latency covering EVERY
+# model, including the two Ollama ones (ornith:35b, qwen2.5:3b-instruct) that
+# expose no metrics endpoint at all. Series it emits, per
+# internal/observability/prometheus.go:
+#   llm_ttft_seconds                    histogram  time-to-first-token
+#   llm_itl_seconds                     histogram  inter-token latency
+#   llm_tokens_total                    counter    SSE chunks generated
+#   queue_full_requests_total           counter    backpressure rejects
+#   warmup_requests_delayed_total       counter    vLLM slot contention
+#   vllm_prefix_cache_hit_tokens_total  counter    prefix-cache hits
+#   request_duration_seconds            histogram  full request
+#
+# Caveat: these are derived from the SSE byte stream, not from the engine.
+# TTFT is request-start to first byte, ITL is the gap between SSE writes
+# carrying "data: ", and llm_tokens_total counts SSE chunks rather than true
+# tokens. They are also streaming-only — non-streaming requests record
+# nothing. Treat vLLM's own histograms as authoritative for `reasoning`.
+#
+# The Deployment already carried prometheus.io/scrape annotations, but
+# kube-prometheus-stack ignores pod annotations (no additionalScrapeConfigs,
+# serviceMonitorSelectorNilUsesHelmValues: false), so none of the above was
+# ever collected. This ServiceMonitor is what actually wires it up.
 apiVersion: monitoring.coreos.com/v1
 kind: ServiceMonitor
 metadata:
@@ -22,27 +44,3 @@ spec:
       path: /metrics
       scheme: http
 
-      # The gateway is the only source of per-request LLM latency that covers
-      # EVERY model, including the two Ollama ones that expose no metrics at
-      # all. Series it produces (internal/observability/prometheus.go):
-      #   llm_ttft_seconds              histogram  time-to-first-token
-      #   llm_itl_seconds               histogram  inter-token latency
-      #   llm_tokens_total              counter    tokens generated
-      #   queue_full_requests_total     counter    backpressure rejects
-      #   warmup_requests_delayed_total counter    vLLM slot contention
-      #   vllm_prefix_cache_hit_tokens_total counter prefix-cache hits
-      #   request_duration_seconds      histogram  full request
-      #
-      # The Deployment already carried prometheus.io/scrape annotations, but
-      # kube-prometheus-stack ignores pod annotations (no additionalScrape-
-      # Configs, serviceMonitorSelectorNilUsesHelmValues: false) — so none of
-      # the above was being collected. This ServiceMonitor is what actually
-      # wires it up.
-      metricRelabelings:
-        # 3 gateway replicas each build their own in-process histogram. Without
-        # this the label set is otherwise identical, so per-instance series
-        # would collide conceptually in sum()-by aggregations; instance is
-        # retained deliberately so per-replica skew stays visible.
-        - sourceLabels: [__address__]
-          targetLabel: instance
-          replacement: $1
diff --git a/k8s/infra/monitoring/servicemonitors/vllm.yaml b/k8s/infra/monitoring/servicemonitors/vllm.yaml
index dfbd574..e2b0eb9 100644
--- a/k8s/infra/monitoring/servicemonitors/vllm.yaml
+++ b/k8s/infra/monitoring/servicemonitors/vllm.yaml
@@ -36,11 +36,14 @@ spec:
       path: /metrics
       scheme: http
       relabelings:
-        - sourceLabels: [__meta_kubernetes_namespace]
-          targetLabel: namespace
-        - sourceLabels: [__meta_kubernetes_pod_name]
-          targetLabel: pod
-        - sourceLabels: [__meta_kubernetes_service_name]
-          targetLabel: service
-        - sourceLabels: [__meta_kubernetes_pod_label_app_kubernetes_io_name]
+        # Prometheus's Kubernetes service discovery already supplies the
+        # namespace, pod and service labels; the only one that needs deriving
+        # is `app`, since the TEI panels aggregate by it (by (app)) and SD does
+        # not promote arbitrary pod labels.
+        #
+        # `action: replace` is written explicitly. It is the API server's
+        # default for relabeling rules, so omitting it leaves the live object
+        # diverged from this manifest and ArgoCD reports permanent OutOfSync.
+        - action: replace
+          sourceLabels: [__meta_kubernetes_pod_label_app_kubernetes_io_name]
           targetLabel: app