diff --git a/apps/litellm/base/deployment.yaml b/apps/litellm/base/deployment.yaml
index e1a76ed..e0c69d4 100644
--- a/apps/litellm/base/deployment.yaml
+++ b/apps/litellm/base/deployment.yaml
@@ -17,7 +17,15 @@ spec:
     spec:
       containers:
         - name: litellm
-          image: registry.nxtgauge.com/litellm:latest
+          # This app was never included in Flux's root kustomization (see
+          # clusters/production/kustomization.yaml), so it was only ever
+          # deployed by a one-off manual `kubectl apply` and has since
+          # drifted from this file. Corrected to match what's actually
+          # running live (the real upstream image) rather than
+          # registry.nxtgauge.com/litellm:latest, which doesn't appear to
+          # exist/be maintained — using it would have broken a working
+          # deployment the moment this file was wired back into GitOps.
+          image: ghcr.io/berriai/litellm:latest
           command:
             - "/bin/bash"
           args:
diff --git a/apps/ollama/base/deployment.yaml b/apps/ollama/base/deployment.yaml
index c29b291..f1c8b37 100644
--- a/apps/ollama/base/deployment.yaml
+++ b/apps/ollama/base/deployment.yaml
@@ -24,16 +24,30 @@ spec:
           env:
             - name: OLLAMA_HOST
               value: "0.0.0.0:11434"
+            # Keep a loaded model resident for 30 min of inactivity instead
+            # of Ollama's 5-minute default — job-description/resume/cover-
+            # letter traffic is bursty, and reloading a 2.5-5GB model from
+            # disk on every request would add multi-second latency to each
+            # first call after a gap.
+            - name: OLLAMA_KEEP_ALIVE
+              value: "30m"
           volumeMounts:
             - name: ollama-models
               mountPath: /root/.ollama
           resources:
             requests:
-              cpu: 500m
-              memory: 700Mi
-            limits:
               cpu: 1000m
-              memory: 1500Mi
+              memory: 3Gi
+            limits:
+              # qwen3:4b (~2.5GB on disk) and qwen3:8b (~5.2GB) are already
+              # pulled onto the PVC, but the previous 1500Mi limit could
+              # only ever load gemma3:270m — which is why every LiteLLM
+              # model alias was mapped to gemma3:270m regardless of name
+              # (see apps/litellm/base/configmap.yaml). Sized to comfortably
+              # hold qwen3:8b plus KV cache/runtime overhead, with headroom;
+              # node has 16GB total and was at ~26% memory use.
+              cpu: 4000m
+              memory: 8Gi
       volumes:
         - name: ollama-models
           persistentVolumeClaim:
diff --git a/clusters/production/kustomization.yaml b/clusters/production/kustomization.yaml
index 5fea507..d70e560 100644
--- a/clusters/production/kustomization.yaml
+++ b/clusters/production/kustomization.yaml
@@ -8,6 +8,7 @@ resources:
   - ../../apps/nxtgauge-ai-assistant/overlays/prod
   - ../../apps/github-actions-runners/base
   - ../../apps/ollama/base
+  - ../../apps/litellm/base
   - ../../ops/openobserve-alerts
   - flux-system/traceworks2026-image-automation.yaml
   - ../../apps/traceworks2026/overlays/prod