diff --git a/src/common/snippets/src/generator.cpp b/src/common/snippets/src/generator.cpp
index 6a00a95f71a5a1..5db1670022bb7f 100644
--- a/src/common/snippets/src/generator.cpp
+++ b/src/common/snippets/src/generator.cpp
@@ -10,6 +10,7 @@
 #include "snippets/lowered/pass/cleanup_loop_offsets.hpp"
 #include "snippets/lowered/pass/insert_specific_iterations.hpp"
 #include "snippets/lowered/pass/optimize_loop_single_evaluation.hpp"
+#include "snippets/lowered/pass/serialize_control_flow.hpp"
 #include "snippets/lowered/pass/pass.hpp"
 #include "snippets/op/kernel.hpp"
 
diff --git a/src/common/snippets/src/lowered/pass/insert_specific_iterations.cpp b/src/common/snippets/src/lowered/pass/insert_specific_iterations.cpp
index 6f42a5fe035acf..ae5574c7384e1a 100644
--- a/src/common/snippets/src/lowered/pass/insert_specific_iterations.cpp
+++ b/src/common/snippets/src/lowered/pass/insert_specific_iterations.cpp
@@ -125,6 +125,9 @@ bool InsertSpecificIterations::run(LinearIR& linear_ir) {
                 }
                 handlers[LoopInfo::LAST_ITER].run(linear_ir, main_body_begin_it, main_body_end_it);
                 update_loop_params(loop_end, tail_size, tail_size, false);
+            } else if (specific_first_iteration) {
+                handlers[LoopInfo::MAIN_BODY].run(linear_ir, main_body_begin_it, main_body_end_it);
+                update_loop_params(loop_end, work_amount - increment, increment, false);
             }
         }
         modified = true;
diff --git a/src/plugins/intel_cpu/src/transformations/snippets/x64/pass/lowered/brgemm_blocking.cpp b/src/plugins/intel_cpu/src/transformations/snippets/x64/pass/lowered/brgemm_blocking.cpp
index e6ff79cbe4c68f..ab999b5425eb93 100644
--- a/src/plugins/intel_cpu/src/transformations/snippets/x64/pass/lowered/brgemm_blocking.cpp
+++ b/src/plugins/intel_cpu/src/transformations/snippets/x64/pass/lowered/brgemm_blocking.cpp
@@ -81,9 +81,12 @@ bool BrgemmBlocking::run(LinearIR& linear_ir) {
 
         auto apply_m_blocking = [&]() {
             const auto& m = *(out_preordered_dims.rbegin() + 1);
-            auto block_size_m = brgemm->get_m_block_size();
-            if (block_size_m > m)
-                block_size_m = m;
+            const auto block_size_m = brgemm->get_m_block_size();
+            if (block_size_m >= m) {
+                *(in_0_subtensor.rbegin() + 1) = m;
+                *(out_subtensor.rbegin() + 1) = m;
+                return;
+            }
 
             *(in_0_subtensor.rbegin() + 1) = block_size_m;
             *(out_subtensor.rbegin() + 1) = block_size_m;
@@ -102,9 +105,12 @@ bool BrgemmBlocking::run(LinearIR& linear_ir) {
 
         auto apply_n_blocking = [&]() {
             const auto& n = *out_preordered_dims.rbegin();
-            auto block_size_n = brgemm->get_n_block_size();
-            if (block_size_n > n)
-                block_size_n = n;
+            const auto block_size_n = brgemm->get_n_block_size();
+            if (block_size_n >= n) {
+                *in_1_subtensor.rbegin() = n;
+                *out_subtensor.rbegin() = n;
+                return;
+            }
 
             *in_1_subtensor.rbegin() = block_size_n;
             *out_subtensor.rbegin() = block_size_n;
@@ -124,15 +130,19 @@ bool BrgemmBlocking::run(LinearIR& linear_ir) {
         auto apply_k_blocking = [&]() {
             const auto& k = *in_0_planar_dims.rbegin();
             OPENVINO_ASSERT(k == *(in_1_planar_dims.rbegin() + 1), "Brgemm input descriptors have different K dimension value.");
-            auto block_size_k = brgemm->get_k_block_size();
-            if (block_size_k > k)
-                block_size_k = k;
+            const auto block_size_k = brgemm->get_k_block_size();
+            if (block_size_k >= k) {
+                *in_0_subtensor.rbegin() = k;
+                *(in_1_subtensor.rbegin() + 1) = k;
+                brgemm->set_beta(0.f);
+                return;
+            }
 
             *in_0_subtensor.rbegin() = block_size_k;
             *(in_1_subtensor.rbegin() + 1) = block_size_k;
             auto loop_begin_it = expr_it, loop_end_it = std::next(expr_it);
             std::vector<LoopPort> entries{LoopPort(brgemm_expr->get_input_port(0), true, 0),
-                                            LoopPort(brgemm_expr->get_input_port(1), true, 1)};
+                                          LoopPort(brgemm_expr->get_input_port(1), true, 1)};
             if (brgemm->is_with_compensations()) {
                 entries.emplace_back(brgemm_expr->get_input_port(2), false, 1);
             } else if (brgemm->is_amx()) {