diff --git a/helm/kagent/templates/modelconfig.yaml b/helm/kagent/templates/modelconfig.yaml
index 7004aeec03..3a806a75fc 100644
--- a/helm/kagent/templates/modelconfig.yaml
+++ b/helm/kagent/templates/modelconfig.yaml
@@ -21,6 +21,9 @@ spec:
   {{- with $model }}
   provider: {{ .provider | quote }}
   model: {{ .model | quote }}
+  {{- if hasKey $model "stream" }}
+  stream: {{ $model.stream }}
+  {{- end }}
   {{- if $model.apiKeySecretRef }}
   apiKeySecret: {{.apiKeySecretRef}}
   {{- end }}
diff --git a/helm/kagent/tests/modelconfig_test.yaml b/helm/kagent/tests/modelconfig_test.yaml
index 1362c086dc..f5338001a2 100644
--- a/helm/kagent/tests/modelconfig_test.yaml
+++ b/helm/kagent/tests/modelconfig_test.yaml
@@ -2,6 +2,36 @@ suite: test modelconfig
 templates:
   - modelconfig.yaml
 tests:
+  - it: should preserve the API streaming default when unspecified
+    asserts:
+      - notExists:
+          path: spec.stream
+
+  - it: should preserve an explicit false streaming option
+    set:
+      providers.openAI.stream: false
+    asserts:
+      - equal:
+          path: spec.stream
+          value: false
+
+  - it: should preserve an explicit true streaming option
+    set:
+      providers.openAI.stream: true
+    asserts:
+      - equal:
+          path: spec.stream
+          value: true
+
+  - it: should use the selected provider streaming option
+    set:
+      providers.default: anthropic
+      providers.anthropic.stream: false
+    asserts:
+      - equal:
+          path: spec.stream
+          value: false
+
   - it: should render modelconfig with default provider
     asserts:
       - hasDocuments:
diff --git a/helm/kagent/values.yaml b/helm/kagent/values.yaml
index 7d4b2adedc..023e644638 100644
--- a/helm/kagent/values.yaml
+++ b/helm/kagent/values.yaml
@@ -640,6 +640,8 @@ ui:
 # https://kagent.dev/docs/getting-started/configuring-providers
 
 providers:
+  # Each provider can set `stream: false` to disable LLM response streaming.
+  # When omitted, ModelConfig uses its API default (true).
   default: openAI
   # -- Annotations added to the metadata of the generated default ModelConfig
   # (the one derived from `providers.default`). Omitted from the resource when empty.