diff --git a/apps/base/customer1/openclaw/a100-vllm.yaml b/apps/base/customer1/openclaw/a100-vllm.yaml
index 5e565ef..3362eda 100644
--- a/apps/base/customer1/openclaw/a100-vllm.yaml
+++ b/apps/base/customer1/openclaw/a100-vllm.yaml
@@ -32,7 +32,7 @@ spec:
             - name: TOKENIZER_MODE
               value: "hf"
           args:
-            - --model=llmfan46/Qwen3.5-27B-uncensored-heretic
+            - --model=Youssofal/Qwen3.6-27B-Abliterated-Heretic-Uncensored-BF16
             - --host=0.0.0.0
             - --port=8000
             - --tensor-parallel-size=1
@@ -53,11 +53,11 @@ spec:
           resources:
             limits:
               nvidia.com/gpu: 1
-              memory: "80Gi"
+              memory: "100Gi"
               cpu: "16"
             requests:
               nvidia.com/gpu: 1
-              memory: "60Gi"
+              memory: "80Gi"
               cpu: "8"
           volumeMounts:
             - name: model-cache
@@ -65,12 +65,12 @@ spec:
       volumes:
         - name: model-cache
           persistentVolumeClaim:
-            claimName: vllm-model-qwen3.5-27b-uncensored
+            claimName: vllm-model-qwen3.6-27b-uncensored
 ---
 apiVersion: v1
 kind: PersistentVolumeClaim
 metadata:
-  name: vllm-model-qwen3.5-27b-uncensored
+  name: vllm-model-qwen3.6-27b-uncensored
   namespace: customer1
 spec:
   accessModes: