diff --git a/tests/integration/defs/accuracy/accuracy_core.py b/tests/integration/defs/accuracy/accuracy_core.py
index e30c6e2c2cb4..d7b885ad742f 100644
--- a/tests/integration/defs/accuracy/accuracy_core.py
+++ b/tests/integration/defs/accuracy/accuracy_core.py
@@ -452,7 +452,7 @@ class LongBenchV2(AccuracyTask):
     EVALUATOR_KWARGS = dict(
         dataset_path=DATASET_DIR,
         length="medium",
-        max_len=120000,
+        max_input_length=120000,
         apply_chat_template=True,
         random_seed=0,
     )
diff --git a/tests/integration/defs/accuracy/test_llm_api_pytorch.py b/tests/integration/defs/accuracy/test_llm_api_pytorch.py
index fc4ba2952090..db6ef3c4e8e4 100644
--- a/tests/integration/defs/accuracy/test_llm_api_pytorch.py
+++ b/tests/integration/defs/accuracy/test_llm_api_pytorch.py
@@ -1118,6 +1118,7 @@ def test_fp8_prequantized(self):
             task = MMLU(self.MODEL_NAME)
             task.evaluate(llm)
 
+    @skip_pre_hopper
     def test_fp8_vswa_reuse(self):
         # NOTE: Test with VSWA kv cache config.
         kv_cache_config = KvCacheConfig(
@@ -1132,6 +1133,7 @@ def test_fp8_vswa_reuse(self):
             task = MMLU(self.MODEL_NAME)
             task.evaluate(llm)
 
+    @skip_pre_hopper
     @pytest.mark.parametrize("backend", ["xgrammar"])
     def test_fp8_guided_decoding_vswa_reuse(self, backend: str, mocker):
         mocker.patch.dict(os.environ, {"TRTLLM_XGUIDANCE_LENIENT": "1"})
@@ -5276,6 +5278,7 @@ class TestNemotronV3Nano(LlmapiAccuracyTestHarness):
     EXTRA_EVALUATOR_KWARGS = dict(chat_template_kwargs=dict(
         enable_thinking=False))
 
+    @pytest.mark.skip_less_device_memory(80000)
     def test_auto_dtype(self):
         with LLM(
                 f"{llm_models_root()}/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16",