diff --git a/tests/integration/defs/accuracy/references/mmlu.yaml b/tests/integration/defs/accuracy/references/mmlu.yaml
index 86a07220237e..21bb5798ccbc 100644
--- a/tests/integration/defs/accuracy/references/mmlu.yaml
+++ b/tests/integration/defs/accuracy/references/mmlu.yaml
@@ -66,6 +66,9 @@ meta-llama/Llama-3.3-70B-Instruct:
     accuracy: 81.02
 meta-llama/Llama-4-Maverick-17B-128E-Instruct:
   - accuracy: 86.40
+  - quant_algo: FP8
+    kv_cache_quant_algo: FP8
+    accuracy: 0.0
 meta-llama/Llama-4-Scout-17B-16E-Instruct:
   - accuracy: 80.00
   - quant_algo: NVFP4
diff --git a/tests/integration/defs/accuracy/test_llm_api_pytorch.py b/tests/integration/defs/accuracy/test_llm_api_pytorch.py
index 204094787043..23665df35af9 100644
--- a/tests/integration/defs/accuracy/test_llm_api_pytorch.py
+++ b/tests/integration/defs/accuracy/test_llm_api_pytorch.py
@@ -408,6 +408,25 @@ def test_auto_dtype(self, cuda_graph, tp_size, pp_size, ep_size):
             task = GSM8K(self.MODEL_NAME)
             task.evaluate(llm)
 
+    @skip_pre_hopper
+    @pytest.mark.skip_less_device(8)
+    @parametrize_with_ids("cuda_graph", [False, True])
+    @pytest.mark.parametrize("tp_size,pp_size,ep_size", [(8, 1, 1), (8, 1, 4),
+                                                         (8, 1, 8)],
+                             ids=["tp8", "tp8ep4", "tp8ep8"])
+    def test_fp8_prequantized(self, cuda_graph, tp_size, pp_size, ep_size):
+        model_path = f"{llm_models_root()}/llama4-models/nvidia/Llama-4-Maverick-17B-128E-Instruct-FP8"
+        with LLM(model_path,
+                 tensor_parallel_size=tp_size,
+                 pipeline_parallel_size=pp_size,
+                 moe_expert_parallel_size=ep_size,
+                 cuda_graph_config=CudaGraphConfig()
+                 if cuda_graph else None) as llm:
+            task = MMLU(self.MODEL_NAME)
+            task.evaluate(llm)
+            task = GSM8K(self.MODEL_NAME)
+            task.evaluate(llm)
+
 
 class TestLlama4ScoutInstruct(LlmapiAccuracyTestHarness):
     MODEL_NAME = "meta-llama/Llama-4-Scout-17B-16E-Instruct"
diff --git a/tests/integration/test_lists/qa/llm_sanity_test.txt b/tests/integration/test_lists/qa/llm_sanity_test.txt
index 4c01e492e1b9..2cfda934e59a 100644
--- a/tests/integration/test_lists/qa/llm_sanity_test.txt
+++ b/tests/integration/test_lists/qa/llm_sanity_test.txt
@@ -35,6 +35,9 @@ accuracy/test_llm_api_pytorch.py::TestLlama3_3_70BInstruct::test_nvfp4_tp4
 accuracy/test_llm_api_pytorch.py::TestLlama4MaverickInstruct::test_auto_dtype[tp8-cuda_graph=False]
 accuracy/test_llm_api_pytorch.py::TestLlama4MaverickInstruct::test_auto_dtype[tp8ep4-cuda_graph=True]
 accuracy/test_llm_api_pytorch.py::TestLlama4MaverickInstruct::test_auto_dtype[tp8ep8-cuda_graph=True]
+accuracy/test_llm_api_pytorch.py::TestLlama4MaverickInstruct::test_fp8_prequantized[tp8-cuda_graph=False]
+accuracy/test_cli_flow.py::TestLlama4MaverickInstruct::test_auto_dtype[tp8-cuda_graph=False]
+accuracy/test_cli_flow.py::TestLlama4MaverickInstruct::test_fp8_prequantized[tp8-cuda_graph=False]
 accuracy/test_llm_api_pytorch.py::TestLlama4ScoutInstruct::test_auto_dtype[tp8-cuda_graph=False]
 accuracy/test_llm_api_pytorch.py::TestLlama4ScoutInstruct::test_auto_dtype[tp8ep4-cuda_graph=True]
 accuracy/test_llm_api_pytorch.py::TestLlama4ScoutInstruct::test_auto_dtype[tp8ep8-cuda_graph=True]