diff --git a/tools/mlir_bench/mlp_bench.sh b/tools/mlir_bench/mlp_bench.sh
index fb0d9866c2a61a..71ce464973c596 100755
--- a/tools/mlir_bench/mlp_bench.sh
+++ b/tools/mlir_bench/mlp_bench.sh
@@ -53,6 +53,10 @@ if ! [ "$(command -v ${BENCH_RUNNER})" ]; then
   echo "Missing benchmark runner"
   exit 1
 fi
+if [ "${BASELINE_MODEL}" ] && [ ${IS_DYNAMIC} ]; then
+  echo "Baseline models with dynamic shapes not supported"
+  exit 1
+fi
 
 # Kernel config.
 INPUT_SIZES=( 1024 2048 4096 8192 )
@@ -86,8 +90,11 @@ for OUT_SIZE in "${OUTPUT_SIZES[@]}"; do
         # No native support for bf16, use simple f16 instead.
         PRECISION="f16"
     fi
-    BENCH_FLAGS="-m ${MODEL_NAME} -d CPU -data_shape [${OUT_SIZE,IN_SIZE}]\
-        -ip ${PRECISION}"
+    if [ ${IS_DYNAMIC} ]; then
+        DATA_SHAPE=(-data_shape [${OUT_SIZE,IN_SIZE}])
+    fi
+    BENCH_FLAGS="-m ${MODEL_NAME} -d CPU \
+        -ip ${PRECISION} ${DATA_SHAPE[@]}"
     ${BENCH_RUNNER} ${BENCH_FLAGS} 2>/dev/null | \
         sed -nE "s/.*\[ INFO \]\s*Median:\s*([0-9.]+).*/\\1/p"
   done
diff --git a/tools/mlir_bench/ov_model_gen.py b/tools/mlir_bench/ov_model_gen.py
index 7ded74a3c92dc4..087023cb222356 100644
--- a/tools/mlir_bench/ov_model_gen.py
+++ b/tools/mlir_bench/ov_model_gen.py
@@ -184,12 +184,14 @@ def forward(self, a, b):
         return self.relu(c)
 
 
-def baseline_MLP(sizes: list[int], data_type: str) -> tuple[nn.Model, list]:
+def baseline_MLP(model_desc: str, data_type: str, is_dynamic: bool) -> tuple[nn.Model, list]:
+    sizes = get_layer_sizes(model_desc)
     assert len(sizes) == 3, "Invalid baseline MLP sizes"
     mlp = BaselineMLP(sizes, get_torch_type(data_type))
-    m = sizes[0]
-    n = sizes[1]
-    k = sizes[2]
+    input_shapes = get_layer_inputs(model_desc, is_dynamic)[0]
+    m = input_shapes[0]
+    n = input_shapes[1]
+    k = input_shapes[2]
     ov_type = get_ov_type(data_type)
     inputs = [(ov.PartialShape([m, k]), ov_type), (ov.PartialShape([k, n]), ov_type)]
     return (mlp, inputs)
@@ -197,10 +199,9 @@ def baseline_MLP(sizes: list[int], data_type: str) -> tuple[nn.Model, list]:
 
 def generate_baseline_model(model_desc: str, data_type: str, shape_type: str, file_name: str):
     model_name = get_layer_name(model_desc)
-    input_shapes = get_layer_inputs(model_desc, shape_type == 'dynamic')
 
     if model_name == 'mlp':
-        baseline_tuple = baseline_MLP(*input_shapes, data_type)
+        baseline_tuple = baseline_MLP(model_desc, data_type, shape_type == 'dynamic')
     else:
         assert False, f"Unsupported baseline model data type {model_name}"