diff --git a/tests/integration/defs/perf/pytorch_model_config.py b/tests/integration/defs/perf/pytorch_model_config.py
index 5851ea3fce..8073d1db93 100644
--- a/tests/integration/defs/perf/pytorch_model_config.py
+++ b/tests/integration/defs/perf/pytorch_model_config.py
@@ -32,42 +32,66 @@ def get_model_yaml_config(model_label: str) -> dict:
         'use_cuda_graph': True,
         'cuda_graph_padding_enabled': True,
     }
-    model_configs = {
-        'deepseek_r1-bench-pytorch-float16-maxbs:1-maxnt:8192-input_output_len:1000,2000-quant:fp8-reqs:10-ep:4-gpus:8':
+
+    # Pattern-based configurations for models matching specific substrings
+    # This allows for flexible configuration of models based on naming patterns
+    pattern_configs = [
+        # DeepSeek R1 models with MTP speculative decoding
         {
-            'use_cuda_graph': True,
-            'speculative_config': {
-                'decoding_type': 'MTP',
-                'num_nextn_predict_layers': 3
+            'patterns': [
+                'deepseek_r1-bench-pytorch-float16-maxbs:1-maxnt:8192-input_output_len:1000,2000-quant:fp8-reqs:10-ep:4-gpus:8',
+                'deepseek_r1_nvfp4-bench-pytorch-float16-maxbs:1-maxnt:8192-input_output_len:1000,2000-quant:nvfp4-reqs:10-ep:4-tp:8-gpus:8'
+            ],
+            'config': {
+                'use_cuda_graph': True,
+                'speculative_config': {
+                    'decoding_type': 'MTP',
+                    'num_nextn_predict_layers': 3
+                }
             }
         },
-        'deepseek_r1_nvfp4-bench-pytorch-float16-maxbs:1-maxnt:8192-input_output_len:1000,2000-quant:nvfp4-reqs:10-ep:4-tp:8-gpus:8':
+        # DeepSeek R1 models with large batch sizes and cuda graph padding
         {
-            'use_cuda_graph': True,
-            'speculative_config': {
-                'decoding_type': 'MTP',
-                'num_nextn_predict_layers': 3
+            'patterns': [
+                'deepseek_r1-bench-pytorch-float16-maxbs:384-maxnt:1536-input_output_len:1000,2000-quant:nvfp4-reqs:49152-con:3072-ep:8-gpus:8',
+                'deepseek_r1_nvfp4-bench-pytorch-float16-maxbs:384-maxnt:1536-input_output_len:1000,2000-quant:nvfp4-reqs:49152-con:3072-ep:8-gpus:8'
+            ],
+            'config': {
+                'cuda_graph_padding_enabled': True,
+                'cuda_graph_batch_sizes':
+                [1, 2, 4, 8, 16, 32, 64, 128, 256, 384]
             }
         },
-        'deepseek_r1-bench-pytorch-float16-maxbs:128-maxnt:1127-input_output_len:1000,2000-quant:fp8-reqs:5120-con:1024-ep:8-gpus:8':
+        # DeepSeek R1 model with specific batch size 128
         {
-            'cuda_graph_batch_sizes': [128]
+            'patterns':
+            'deepseek_r1-bench-pytorch-float16-maxbs:128-maxnt:1127-input_output_len:1000,2000-quant:fp8-reqs:5120-con:1024-ep:8-gpus:8',
+            'config': {
+                'cuda_graph_batch_sizes': [128]
+            }
         },
-        'deepseek_r1-bench-pytorch-float16-maxbs:384-maxnt:1536-input_output_len:1000,2000-quant:nvfp4-reqs:49152-con:3072-ep:8-gpus:8':
+        # Llama Nemotron models with attention_dp disabled to prevent hangs
         {
-            'cuda_graph_padding_enabled': True,
-            'cuda_graph_batch_sizes': [1, 2, 4, 8, 16, 32, 64, 128, 256, 384]
+            'patterns': [
+                'llama_v3.1_nemotron_ultra_253b_fp8-bench-pytorch-float8',
+                'llama_v3.3_nemotron_super_49b_fp8-bench-pytorch-float8',
+                'llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16'
+            ],
+            'config': {
+                # True causes hang, needs model-specific fix.
+                'enable_attention_dp': False,
+            }
         },
-        'deepseek_r1_nvfp4-bench-pytorch-float16-maxbs:384-maxnt:1536-input_output_len:1000,2000-quant:nvfp4-reqs:49152-con:3072-ep:8-gpus:8':
-        {
-            'cuda_graph_padding_enabled': True,
-            'cuda_graph_batch_sizes': [1, 2, 4, 8, 16, 32, 64, 128, 256, 384]
-        }
-    }
-    # get model name from model_label
-    model_name = next(
-        (key for key in model_configs if key in model_label.lower()), None)
-    if model_name:
-        base_config.update(model_configs[model_name])
+    ]
+
+    # Apply pattern-based configurations on top of base config
+    for pattern_config in pattern_configs:
+        patterns = pattern_config['patterns']
+        if isinstance(patterns, str):
+            patterns = [patterns]
+        for pattern in patterns:
+            if pattern in model_label.lower():
+                base_config.update(pattern_config['config'])
+                break  # Stop checking other patterns for this config once we find a match
 
     return base_config
diff --git a/tests/integration/defs/perf/test_perf.py b/tests/integration/defs/perf/test_perf.py
index 96dd9a05c7..5f283aec71 100644
--- a/tests/integration/defs/perf/test_perf.py
+++ b/tests/integration/defs/perf/test_perf.py
@@ -57,10 +57,13 @@ MODEL_PATH_DICT = {
     "llama_v3.1_70b_instruct": "llama-3.1-model/Meta-Llama-3.1-70B-Instruct",
     "llama_v3.2_1b": "llama-3.2-models/Llama-3.2-1B",
     "llama_v3.1_nemotron_nano_8b": "Llama-3.1-Nemotron-Nano-8B-v1",
+    "llama_v3.1_nemotron_nano_8b_fp8": "Llama-3.1-Nemotron-Nano-8B-v1-FP8",
     "llama_v3.3_nemotron_super_49b":
     "nemotron-nas/Llama-3_3-Nemotron-Super-49B-v1",
-    "llama_v3.1_nemotron_ultra_253b":
-    "nemotron-nas/Llama-3_1-Nemotron-Ultra-253B-v1",
+    "llama_v3.3_nemotron_super_49b_fp8":
+    "nemotron-nas/Llama-3_3-Nemotron-Super-49B-v1-FP8",
+    "llama_v3.1_nemotron_ultra_253b_fp8":
+    "nemotron-nas/Llama-3_1-Nemotron-Ultra-253B-v1-FP8",
     # "llama_30b": "llama-models/llama-30b-hf",
     "mixtral_8x7b_v0.1": "Mixtral-8x7B-v0.1",
     "mixtral_8x7b_v0.1_instruct": "Mixtral-8x7B-Instruct-v0.1",
@@ -107,10 +110,14 @@ HF_MODEL_PATH = {
     "llama_v3.1_70b_hf": "meta-llama/Llama-3.1-70B",
     "llama_v3.1_405b_hf": "meta-llama/Llama-3.1-405B",
     "llama_v3.1_nemotron_nano_8b_hf": "nvidia/Llama-3.1-Nemotron-Nano-8B-v1",
+    "llama_v3.1_nemotron_nano_8b_fp8_hf":
+    "nvidia/Llama-3.1-Nemotron-Nano-8B-v1-FP8",
     "llama_v3.3_nemotron_super_49b_hf":
     "nvidia/Llama-3_3-Nemotron-Super-49B-v1",
-    "llama_v3.1_nemotron_ultra_253b_hf":
-    "nvidia/Llama-3_1-Nemotron-Ultra-253B-v1",
+    "llama_v3.3_nemotron_super_49b_fp8_hf":
+    "nvidia/Llama-3_3-Nemotron-Super-49B-v1-FP8",
+    "llama_v3.1_nemotron_ultra_253b_fp8_hf":
+    "nvidia/Llama-3_1-Nemotron-Ultra-253B-v1-FP8",
     "mixtral_8x7b_v0.1_hf": "mistralai/Mixtral-8x7B-v0.1",
     "mixtral_8x7b_v0.1_instruct_hf": "mistralai/Mixtral-8x7B-Instruct-v0.1",
     "mistral_7b_v0.1_hf": "mistralai/Mistral-7B-v0.1",
@@ -125,7 +132,10 @@ LORA_MODEL_PATH = {
 TIMING_CACHE_DIR = os.environ.get("TIMING_CACHE_DIR", "")
 
 TRUST_REMOTE_CODE_MODELS = {  # these models require explicit trust_remote_code=True
-    "llama_v3.3_nemotron_super_49b"
+    "llama_v3.3_nemotron_super_49b",
+    "llama_v3.3_nemotron_super_49b_fp8",
+    "llama_v3.1_nemotron_ultra_253b",
+    "llama_v3.1_nemotron_ultra_253b_fp8",
 }
 
 
diff --git a/tests/integration/test_lists/qa/trt_llm_release_perf_test.yml b/tests/integration/test_lists/qa/trt_llm_release_perf_test.yml
index b99bba9969..21374329b4 100644
--- a/tests/integration/test_lists/qa/trt_llm_release_perf_test.yml
+++ b/tests/integration/test_lists/qa/trt_llm_release_perf_test.yml
@@ -90,6 +90,15 @@ trt_llm_release_perf_test:
   - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b-bench-pytorch-bfloat16-maxbs:512-input_output_len:1000,1000-reqs:8-con:1]
   - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b-bench-pytorch-bfloat16-maxbs:512-maxnt:20000-input_output_len:20000,2000-reqs:8-con:1]
   - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b-bench-pytorch-bfloat16-maxbs:512-input_output_len:500,2000-reqs:500-con:250]
+  # FP8 prequantized pyt backend
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b_fp8-bench-pytorch-float8-maxbs:512-maxnt:5000-input_output_len:5000,500-reqs:8-con:1]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b_fp8-bench-pytorch-float8-maxbs:512-input_output_len:500,2000-reqs:8-con:1]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b_fp8-bench-pytorch-float8-maxbs:512-input_output_len:1000,1000-reqs:8-con:1]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b_fp8-bench-pytorch-float8-maxbs:512-maxnt:20000-input_output_len:20000,2000-reqs:8-con:1]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b_fp8-bench-pytorch-float8-maxbs:512-maxnt:5000-input_output_len:5000,500-reqs:500-con:250]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b_fp8-bench-pytorch-float8-maxbs:512-input_output_len:500,2000-reqs:500-con:250]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b_fp8-bench-pytorch-float8-maxbs:512-input_output_len:1000,1000-reqs:500-con:250]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b_fp8-bench-pytorch-float8-maxbs:512-maxnt:20000-input_output_len:20000,2000-reqs:500-con:250]
   #long time llama_nemotron cases
   - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b-bench-bfloat16-maxbs:64-input_output_len:5000,500-reqs:8-con:1] # timeout for l20, l40s, a100
   - perf/test_perf.py::test_perf[llama_v3.1_nemotron_nano_8b-bench-bfloat16-maxbs:64-input_output_len:5000,500-quant:fp8-reqs:8-con:1] #timeout for l20, l40s, failed for a100
@@ -276,6 +285,8 @@ trt_llm_release_perf_test:
   tests:
   - perf/test_perf.py::test_perf[llama_v3.1_70b-bench-bfloat16-input_output_len:512,200-quant:fp8-tp:4]
   - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-float8-input_output_len:128,128-tp:4]
+  # Llama-Nemotron-Super-49B-v3.3
+  # cpp
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-bfloat16-maxbs:64-input_output_len:5000,500-reqs:4-con:1-gpus:4]
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-bfloat16-maxbs:64-input_output_len:5000,500-quant:fp8-reqs:4-con:1-gpus:4]
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-bfloat16-maxbs:64-input_output_len:500,2000-reqs:4-con:1-gpus:4]
@@ -283,6 +294,18 @@ trt_llm_release_perf_test:
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-bfloat16-maxbs:16-input_output_len:5000,500-con:250-gpus:4]
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-bfloat16-maxbs:16-input_output_len:5000,500-quant:fp8-con:250-gpus:4]
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-bfloat16-maxbs:16-input_output_len:500,2000-con:250-gpus:4]
+  # pyt
+  # bfloat16
+  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16-maxbs:1-maxnt:5000-input_output_len:5000,500-reqs:8-con:1-tp:4-gpus:4]
+  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16-maxbs:1-input_output_len:500,2000-reqs:8-con:1-tp:4-gpus:4]
+  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16-maxbs:256-maxnt:5000-input_output_len:5000,500-reqs:250-con:250-tp:4-gpus:4]
+  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16-maxbs:256-input_output_len:500,2000-reqs:250-con:250-tp:4-gpus:4]
+  # fp8 prequantized
+  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b_fp8-bench-pytorch-float8-maxbs:1-maxnt:5000-input_output_len:5000,500-reqs:8-con:1-tp:4-gpus:4]
+  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b_fp8-bench-pytorch-float8-maxbs:1-input_output_len:500,2000-reqs:8-con:1-tp:4-gpus:4]
+  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b_fp8-bench-pytorch-float8-maxbs:256-maxnt:5000-input_output_len:5000,500-reqs:250-con:250-tp:4-gpus:4]
+  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b_fp8-bench-pytorch-float8-maxbs:256-input_output_len:500,2000-reqs:250-con:250-tp:4-gpus:4]
+
 
 - condition:
     ranges:
@@ -331,6 +354,12 @@ trt_llm_release_perf_test:
   - perf/test_perf.py::test_perf[llama_v3.1_nemotron_ultra_253b-bench-bfloat16-maxbs:64-input_output_len:500,2000-quant:fp8-reqs:8-con:1-tp:8-gpus:8]
   - perf/test_perf.py::test_perf[llama_v3.1_nemotron_ultra_253b-bench-bfloat16-maxbs:64-input_output_len:5000,500-quant:fp8-reqs:250-con:250-tp:8-gpus:8]
   - perf/test_perf.py::test_perf[llama_v3.1_nemotron_ultra_253b-bench-bfloat16-maxbs:64-input_output_len:500,2000-quant:fp8-reqs:250-con:250-tp:8-gpus:8]
+  # pyt backend, fp8 pre-quantized
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_ultra_253b_fp8-bench-pytorch-float8-maxbs:1-maxnt:5000-input_output_len:5000,500-reqs:8-con:1-tp:8-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_ultra_253b_fp8-bench-pytorch-float8-maxbs:1-input_output_len:500,2000-reqs:8-con:1-tp:8-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_ultra_253b_fp8-bench-pytorch-float8-maxbs:256-maxnt:5000-input_output_len:5000,500-reqs:250-con:250-tp:8-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.1_nemotron_ultra_253b_fp8-bench-pytorch-float8-maxbs:256-input_output_len:500,2000-reqs:250-con:250-tp:8-gpus:8]
+
 
 - condition:
     ranges: