diff --git a/examples/qualcomm/oss_scripts/albert.py b/examples/qualcomm/oss_scripts/albert.py
index 3be48215ac6..d529e5db734 100644
--- a/examples/qualcomm/oss_scripts/albert.py
+++ b/examples/qualcomm/oss_scripts/albert.py
@@ -30,6 +30,9 @@
 
 
 def main(args):
+    if args.compile_only and args.pre_gen_pte:
+        raise RuntimeError("Cannot set both compile_only and pre_gen_pte as true")
+
     skip_node_id_set, skip_node_op_set = parse_skip_delegation_node(args)
 
     os.makedirs(args.artifact, exist_ok=True)
@@ -60,26 +63,32 @@ def main(args):
     module = AutoModelForMaskedLM.from_pretrained(model_name, config=config).eval()
     pte_filename = "albert_qnn_q16"
 
-    # lower to QNN
-    passes_job = get_capture_program_passes()
-    build_executorch_binary(
-        module,
-        inputs[0],
-        args.model,
-        f"{args.artifact}/{pte_filename}",
-        dataset=inputs,
-        skip_node_id_set=skip_node_id_set,
-        skip_node_op_set=skip_node_op_set,
-        quant_dtype=QuantDtype.use_16a16w,
-        passes_job=passes_job,
-        shared_buffer=args.shared_buffer,
-    )
+    # Skip lowering/compilation if using pre-generated PTE
+    if not args.pre_gen_pte:
+        # lower to QNN
+        passes_job = get_capture_program_passes()
+        build_executorch_binary(
+            module,
+            inputs[0],
+            args.model,
+            f"{args.artifact}/{pte_filename}",
+            dataset=inputs,
+            skip_node_id_set=skip_node_id_set,
+            skip_node_op_set=skip_node_op_set,
+            quant_dtype=QuantDtype.use_16a16w,
+            passes_job=passes_job,
+            shared_buffer=args.shared_buffer,
+        )
 
     if args.compile_only:
         return
 
     workspace = f"/data/local/tmp/{getpass.getuser()}/executorch/{pte_filename}"
-    pte_path = f"{args.artifact}/{pte_filename}.pte"
+    pte_path = (
+        f"{args.pre_gen_pte}/{pte_filename}.pte"
+        if args.pre_gen_pte
+        else f"{args.artifact}/{pte_filename}.pte"
+    )
 
     adb = SimpleADB(
         qnn_sdk=os.getenv("QNN_SDK_ROOT"),
diff --git a/examples/qualcomm/oss_scripts/bert.py b/examples/qualcomm/oss_scripts/bert.py
index 0f9255cefdb..aa41df6ff4d 100644
--- a/examples/qualcomm/oss_scripts/bert.py
+++ b/examples/qualcomm/oss_scripts/bert.py
@@ -30,6 +30,9 @@
 
 
 def main(args):
+    if args.compile_only and args.pre_gen_pte:
+        raise RuntimeError("Cannot set both compile_only and pre_gen_pte as true")
+
     skip_node_id_set, skip_node_op_set = parse_skip_delegation_node(args)
 
     os.makedirs(args.artifact, exist_ok=True)
@@ -57,26 +60,32 @@ def main(args):
     ).eval()
     pte_filename = "bert_qnn_q16"
 
-    # lower to QNN
-    passes_job = get_capture_program_passes()
-    build_executorch_binary(
-        module,
-        inputs[0],
-        args.model,
-        f"{args.artifact}/{pte_filename}",
-        dataset=inputs,
-        skip_node_id_set=skip_node_id_set,
-        skip_node_op_set=skip_node_op_set,
-        quant_dtype=QuantDtype.use_16a8w,
-        passes_job=passes_job,
-        shared_buffer=args.shared_buffer,
-    )
+    # Skip lowering/compilation if using pre-generated PTE
+    if not args.pre_gen_pte:
+        # lower to QNN
+        passes_job = get_capture_program_passes()
+        build_executorch_binary(
+            module,
+            inputs[0],
+            args.model,
+            f"{args.artifact}/{pte_filename}",
+            dataset=inputs,
+            skip_node_id_set=skip_node_id_set,
+            skip_node_op_set=skip_node_op_set,
+            quant_dtype=QuantDtype.use_16a8w,
+            passes_job=passes_job,
+            shared_buffer=args.shared_buffer,
+        )
 
     if args.compile_only:
         return
 
     workspace = f"/data/local/tmp/{getpass.getuser()}/executorch/{pte_filename}"
-    pte_path = f"{args.artifact}/{pte_filename}.pte"
+    pte_path = (
+        f"{args.pre_gen_pte}/{pte_filename}.pte"
+        if args.pre_gen_pte
+        else f"{args.artifact}/{pte_filename}.pte"
+    )
 
     adb = SimpleADB(
         qnn_sdk=os.getenv("QNN_SDK_ROOT"),
diff --git a/examples/qualcomm/oss_scripts/distilbert.py b/examples/qualcomm/oss_scripts/distilbert.py
index 7ca05181645..ce88f61ca5c 100644
--- a/examples/qualcomm/oss_scripts/distilbert.py
+++ b/examples/qualcomm/oss_scripts/distilbert.py
@@ -31,6 +31,9 @@
 
 
 def main(args):
+    if args.compile_only and args.pre_gen_pte:
+        raise RuntimeError("Cannot set both compile_only and pre_gen_pte as true")
+
     skip_node_id_set, skip_node_op_set = parse_skip_delegation_node(args)
 
     os.makedirs(args.artifact, exist_ok=True)
@@ -58,26 +61,32 @@ def main(args):
     ).eval()
     pte_filename = "distilbert_qnn_q16"
 
-    # lower to QNN
-    passes_job = get_capture_program_passes()
-    build_executorch_binary(
-        module,
-        inputs[0],
-        args.model,
-        f"{args.artifact}/{pte_filename}",
-        dataset=inputs,
-        skip_node_id_set=skip_node_id_set,
-        skip_node_op_set=skip_node_op_set,
-        quant_dtype=QuantDtype.use_16a8w,
-        passes_job=passes_job,
-        shared_buffer=args.shared_buffer,
-    )
+    # Skip lowering/compilation if using pre-generated PTE
+    if not args.pre_gen_pte:
+        # lower to QNN
+        passes_job = get_capture_program_passes()
+        build_executorch_binary(
+            module,
+            inputs[0],
+            args.model,
+            f"{args.artifact}/{pte_filename}",
+            dataset=inputs,
+            skip_node_id_set=skip_node_id_set,
+            skip_node_op_set=skip_node_op_set,
+            quant_dtype=QuantDtype.use_16a8w,
+            passes_job=passes_job,
+            shared_buffer=args.shared_buffer,
+        )
 
     if args.compile_only:
         return
 
     workspace = f"/data/local/tmp/{getpass.getuser()}/executorch/{pte_filename}"
-    pte_path = f"{args.artifact}/{pte_filename}.pte"
+    pte_path = (
+        f"{args.pre_gen_pte}/{pte_filename}.pte"
+        if args.pre_gen_pte
+        else f"{args.artifact}/{pte_filename}.pte"
+    )
 
     adb = SimpleADB(
         qnn_sdk=os.getenv("QNN_SDK_ROOT"),
diff --git a/examples/qualcomm/oss_scripts/eurobert.py b/examples/qualcomm/oss_scripts/eurobert.py
index a856616bcf2..5e133aed0d1 100644
--- a/examples/qualcomm/oss_scripts/eurobert.py
+++ b/examples/qualcomm/oss_scripts/eurobert.py
@@ -35,6 +35,9 @@
 
 
 def main(args):
+    if args.compile_only and args.pre_gen_pte:
+        raise RuntimeError("Cannot set both compile_only and pre_gen_pte as true")
+
     assert (
         transformers.__version__ >= TRANSFORMERS_VERSION
     ), f"Please ensure transformers version >= {TRANSFORMERS_VERSION}, current version is {transformers.__version__}"
@@ -88,33 +91,40 @@ def replace_rms_norm_with_native_rms_norm(module: torch.nn.Module):
 
     pte_filename = "eurobert_qnn_q16"
 
-    # lower to QNN
-    passes_job = get_capture_program_passes()
-    quantizer = make_quantizer(
-        quant_dtype=QuantDtype.use_16a16w,
-    )
-    quantizer.add_custom_quant_annotations((annotate_eurobert,))
-    with torch.no_grad():
-        build_executorch_binary(
-            model,
-            inputs[0],
-            args.model,
-            f"{args.artifact}/{pte_filename}",
-            dataset=inputs,
-            skip_node_id_set=skip_node_id_set,
-            skip_node_op_set=skip_node_op_set,
-            custom_quantizer=quantizer,
-            passes_job=passes_job,
-            shared_buffer=args.shared_buffer,
+    # Skip lowering/compilation if using pre-generated PTE
+    if not args.pre_gen_pte:
+        # lower to QNN
+        passes_job = get_capture_program_passes()
+        quantizer = make_quantizer(
+            quant_dtype=QuantDtype.use_16a16w,
         )
+        quantizer.add_custom_quant_annotations((annotate_eurobert,))
+        with torch.no_grad():
+            build_executorch_binary(
+                model,
+                inputs[0],
+                args.model,
+                f"{args.artifact}/{pte_filename}",
+                dataset=inputs,
+                skip_node_id_set=skip_node_id_set,
+                skip_node_op_set=skip_node_op_set,
+                custom_quantizer=quantizer,
+                passes_job=passes_job,
+                shared_buffer=args.shared_buffer,
+            )
 
     if args.compile_only:
         return
 
+    pte_path = (
+        f"{args.pre_gen_pte}/{pte_filename}.pte"
+        if args.pre_gen_pte
+        else f"{args.artifact}/{pte_filename}.pte"
+    )
     adb = SimpleADB(
         qnn_sdk=os.getenv("QNN_SDK_ROOT"),
         build_path=f"{args.build_folder}",
-        pte_path=f"{args.artifact}/{pte_filename}.pte",
+        pte_path=pte_path,
         workspace=f"/data/local/tmp/executorch/{pte_filename}",
         device_id=args.device,
         host_id=args.host,
diff --git a/examples/qualcomm/oss_scripts/llama/llama.py b/examples/qualcomm/oss_scripts/llama/llama.py
index 91d82531654..8b54b8e150c 100755
--- a/examples/qualcomm/oss_scripts/llama/llama.py
+++ b/examples/qualcomm/oss_scripts/llama/llama.py
@@ -1023,7 +1023,8 @@ def post_process():
             runner=f"examples/qualcomm/oss_scripts/llama/qnn_llama_runner",
         )
         # No pregen inputs, input_list is not required
-        adb.push(inputs=[], files=[runtime_tokenizer_path])
+        if not args.skip_push:
+            adb.push(inputs=[], files=[runtime_tokenizer_path])
         adb.execute(custom_runner_cmd=runner_cmd)
         adb.pull(output_path=args.artifact, callback=post_process)