diff --git a/microsoft-Phi-3-mini-128k-instruct/QNN/config.json b/microsoft-Phi-3-mini-128k-instruct/QNN/config.json
index 67dede031..8a754eb31 100644
--- a/microsoft-Phi-3-mini-128k-instruct/QNN/config.json
+++ b/microsoft-Phi-3-mini-128k-instruct/QNN/config.json
@@ -60,7 +60,7 @@
         "gs": {
             "type": "GraphSurgeries",
             "surgeries": [
-                { "surgeon": "RemoveRopeMultiCache" },
+                { "surgeon": "RemoveRopeMultiCache", "use_large_cache": true },
                 { "surgeon": "AttentionMaskToSequenceLengths" },
                 { "surgeon": "SimplifiedLayerNormToL2Norm" }
             ],
diff --git a/microsoft-Phi-3.5-mini-instruct/QNN/x2_elite_config.json b/microsoft-Phi-3.5-mini-instruct/QNN/x2_elite_config.json
index 0985fa9da..de0b76c78 100644
--- a/microsoft-Phi-3.5-mini-instruct/QNN/x2_elite_config.json
+++ b/microsoft-Phi-3.5-mini-instruct/QNN/x2_elite_config.json
@@ -64,7 +64,7 @@
         "gs": {
             "type": "GraphSurgeries",
             "surgeries": [
-                { "surgeon": "RemoveRopeMultiCache" },
+                { "surgeon": "RemoveRopeMultiCache", "use_large_cache": true },
                 { "surgeon": "AttentionMaskToSequenceLengths" },
                 { "surgeon": "RemoveGidxFromMatMulNBits" },
                 { "surgeon": "SimplifiedLayerNormToL2Norm" }
diff --git a/microsoft-Phi-3.5-mini-instruct/QNN/x_elite_config.json b/microsoft-Phi-3.5-mini-instruct/QNN/x_elite_config.json
index a09cad46d..d4109b8fb 100644
--- a/microsoft-Phi-3.5-mini-instruct/QNN/x_elite_config.json
+++ b/microsoft-Phi-3.5-mini-instruct/QNN/x_elite_config.json
@@ -64,7 +64,7 @@
         "gs": {
             "type": "GraphSurgeries",
             "surgeries": [
-                { "surgeon": "RemoveRopeMultiCache" },
+                { "surgeon": "RemoveRopeMultiCache", "use_large_cache": true},
                 { "surgeon": "AttentionMaskToSequenceLengths" },
                 { "surgeon": "RemoveGidxFromMatMulNBits" },
                 { "surgeon": "SimplifiedLayerNormToL2Norm" }