diff --git a/tests/basic_correctness/test_cpu_offload.py b/tests/basic_correctness/test_cpu_offload.py
index f0f09ee63c0e6..a5df5639cf948 100644
--- a/tests/basic_correctness/test_cpu_offload.py
+++ b/tests/basic_correctness/test_cpu_offload.py
@@ -1,61 +1,6 @@
-import pytest
-
-from tests.quantization.utils import is_quant_method_supported
-
 from ..utils import compare_two_settings
 
 
 def test_cpu_offload():
     compare_two_settings("meta-llama/Llama-2-7b-hf", [],
                          ["--cpu-offload-gb", "4"])
-
-
-@pytest.mark.skipif(not is_quant_method_supported("fp8"),
-                    reason="fp8 is not supported on this GPU type.")
-def test_cpu_offload_fp8():
-    # Test quantization of an unquantized checkpoint
-    compare_two_settings("meta-llama/Meta-Llama-3-8B-Instruct",
-                         ["--quantization", "fp8"],
-                         ["--quantization", "fp8", "--cpu-offload-gb", "2"])
-    # Test loading a quantized checkpoint
-    compare_two_settings("neuralmagic/Meta-Llama-3-8B-Instruct-FP8", [],
-                         ["--cpu-offload-gb", "2"])
-
-
-@pytest.mark.skipif(not is_quant_method_supported("gptq_marlin"),
-                    reason="gptq_marlin is not supported on this GPU type.")
-def test_cpu_offload_gptq():
-    # Test GPTQ Marlin
-    compare_two_settings("Qwen/Qwen2-1.5B-Instruct-GPTQ-Int4", [],
-                         ["--cpu-offload-gb", "1"])
-    # Test GPTQ
-    compare_two_settings("Qwen/Qwen2-1.5B-Instruct-GPTQ-Int4",
-                         ["--quantization", "gptq"],
-                         ["--quantization", "gptq", "--cpu-offload-gb", "1"])
-
-
-@pytest.mark.skipif(not is_quant_method_supported("awq_marlin"),
-                    reason="awq_marlin is not supported on this GPU type.")
-def test_cpu_offload_awq():
-    # Test AWQ Marlin
-    compare_two_settings("Qwen/Qwen2-1.5B-Instruct-AWQ", [],
-                         ["--cpu-offload-gb", "1"])
-    # Test AWQ
-    compare_two_settings("Qwen/Qwen2-1.5B-Instruct-AWQ",
-                         ["--quantization", "awq"],
-                         ["--quantization", "awq", "--cpu-offload-gb", "1"])
-
-
-@pytest.mark.skipif(not is_quant_method_supported("gptq_marlin"),
-                    reason="gptq_marlin is not supported on this GPU type.")
-def test_cpu_offload_compressed_tensors():
-    # Test wNa16
-    compare_two_settings("nm-testing/tinyllama-oneshot-w4a16-channel-v2", [],
-                         ["--cpu-offload-gb", "1"])
-    # Test w4a16_marlin24
-    compare_two_settings("nm-testing/llama7b-one-shot-2_4-w4a16-marlin24-t",
-                         [], ["--cpu-offload-gb", "1"])
-    # Test w8a8
-    compare_two_settings(
-        "nm-testing/tinyllama-oneshot-w8w8-test-static-shape-change", [],
-        ["--cpu-offload-gb", "1"])
diff --git a/tests/quantization/test_cpu_offload.py b/tests/quantization/test_cpu_offload.py
new file mode 100644
index 0000000000000..6fda503536c06
--- /dev/null
+++ b/tests/quantization/test_cpu_offload.py
@@ -0,0 +1,59 @@
+# Expanded quantized model tests for CPU offloading
+# Base tests: tests/basic_correctness/test_cpu_offload.py
+
+import pytest
+
+from tests.quantization.utils import is_quant_method_supported
+
+from ..utils import compare_two_settings
+
+
+@pytest.mark.skipif(not is_quant_method_supported("fp8"),
+                    reason="fp8 is not supported on this GPU type.")
+def test_cpu_offload_fp8():
+    # Test quantization of an unquantized checkpoint
+    compare_two_settings("meta-llama/Meta-Llama-3-8B-Instruct",
+                         ["--quantization", "fp8"],
+                         ["--quantization", "fp8", "--cpu-offload-gb", "2"])
+    # Test loading a quantized checkpoint
+    compare_two_settings("neuralmagic/Meta-Llama-3-8B-Instruct-FP8", [],
+                         ["--cpu-offload-gb", "2"])
+
+
+@pytest.mark.skipif(not is_quant_method_supported("gptq_marlin"),
+                    reason="gptq_marlin is not supported on this GPU type.")
+def test_cpu_offload_gptq():
+    # Test GPTQ Marlin
+    compare_two_settings("Qwen/Qwen2-1.5B-Instruct-GPTQ-Int4", [],
+                         ["--cpu-offload-gb", "1"])
+    # Test GPTQ
+    compare_two_settings("Qwen/Qwen2-1.5B-Instruct-GPTQ-Int4",
+                         ["--quantization", "gptq"],
+                         ["--quantization", "gptq", "--cpu-offload-gb", "1"])
+
+
+@pytest.mark.skipif(not is_quant_method_supported("awq_marlin"),
+                    reason="awq_marlin is not supported on this GPU type.")
+def test_cpu_offload_awq():
+    # Test AWQ Marlin
+    compare_two_settings("Qwen/Qwen2-1.5B-Instruct-AWQ", [],
+                         ["--cpu-offload-gb", "1"])
+    # Test AWQ
+    compare_two_settings("Qwen/Qwen2-1.5B-Instruct-AWQ",
+                         ["--quantization", "awq"],
+                         ["--quantization", "awq", "--cpu-offload-gb", "1"])
+
+
+@pytest.mark.skipif(not is_quant_method_supported("gptq_marlin"),
+                    reason="gptq_marlin is not supported on this GPU type.")
+def test_cpu_offload_compressed_tensors():
+    # Test wNa16
+    compare_two_settings("nm-testing/tinyllama-oneshot-w4a16-channel-v2", [],
+                         ["--cpu-offload-gb", "1"])
+    # Test w4a16_marlin24
+    compare_two_settings("nm-testing/llama7b-one-shot-2_4-w4a16-marlin24-t",
+                         [], ["--cpu-offload-gb", "1"])
+    # Test w8a8
+    compare_two_settings(
+        "nm-testing/tinyllama-oneshot-w8w8-test-static-shape-change", [],
+        ["--cpu-offload-gb", "1"])