diff --git a/src/core/CL/kernels/CLGEMMLowpOffsetContributionKernel.cpp b/src/core/CL/kernels/CLGEMMLowpOffsetContributionKernel.cpp
index 3888353..d348f2c 100644
--- a/src/core/CL/kernels/CLGEMMLowpOffsetContributionKernel.cpp
+++ b/src/core/CL/kernels/CLGEMMLowpOffsetContributionKernel.cpp
@@ -46,11 +46,18 @@
 
 namespace
 {
-Status validate_arguments(const ITensorInfo *mm_result, const ITensorInfo *vector_sum_col, const ITensorInfo *vector_sum_row,
+Status validate_arguments(const ITensorInfo *mm_result, const ITensorInfo *vector_sum_col, const ITensorInfo *vector_sum_row, const ITensorInfo *bias,
                           int32_t a_offset, int32_t b_offset)
 {
     ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(mm_result, 1, DataType::S32);
 
+    if(bias != nullptr)
+    {
+        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(bias, 1, DataType::S32);
+        ARM_COMPUTE_RETURN_ERROR_ON(bias->num_dimensions() > 1);
+        ARM_COMPUTE_RETURN_ERROR_ON(mm_result->dimension(0) != bias->dimension(0));
+    }
+
     // If a_offset == 0, vector_sum_col can be a nullptr
     if(a_offset != 0)
     {
@@ -64,11 +71,11 @@
         ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(vector_sum_row, 1, DataType::S32);
 
         // Check if input is a 3D reinterpretation
-        const bool reinterpret_as_3d = vector_sum_row != nullptr && mm_result->num_dimensions() > 1 && mm_result->tensor_shape().y() != vector_sum_row->tensor_shape().x();
+        const bool reinterpret_as_3d = mm_result->num_dimensions() > 1 && mm_result->tensor_shape().y() != vector_sum_row->tensor_shape().x();
 
         // Validate input
         ARM_COMPUTE_RETURN_ERROR_ON(reinterpret_as_3d && vector_sum_row->dimension(0) != (mm_result->dimension(1) * mm_result->dimension(2)));
-        ARM_COMPUTE_RETURN_ERROR_ON(!reinterpret_as_3d && vector_sum_row != nullptr && vector_sum_row->dimension(0) != mm_result->dimension(1));
+        ARM_COMPUTE_RETURN_ERROR_ON(!reinterpret_as_3d && vector_sum_row->dimension(0) != mm_result->dimension(1));
 
         TensorShape output_shape = mm_result->tensor_shape();
         if(output_shape.num_dimensions() > 1)
@@ -96,7 +103,7 @@
     return Status{};
 }
 
-std::pair<Status, Window> validate_and_configure_window(ITensorInfo *mm_result, ITensorInfo *vector_sum_col, ITensorInfo *vector_sum_row,
+std::pair<Status, Window> validate_and_configure_window(ITensorInfo *mm_result, ITensorInfo *vector_sum_col, ITensorInfo *vector_sum_row, ITensorInfo *bias,
                                                         int32_t a_offset, int32_t b_offset)
 {
     constexpr unsigned int num_elems_processed_per_iteration = 4;
@@ -119,28 +126,37 @@
         window_changed = window_changed || update_window_and_padding(win, vector_sum_row_access);
     }
 
+    if(bias != nullptr)
+    {
+        AccessWindowStatic bias_access(bias, 0, 0, ceil_to_multiple(bias->dimension(0), num_elems_processed_per_iteration), bias->tensor_shape()[1]);
+        window_changed = window_changed || update_window_and_padding(win, bias_access);
+    }
+
     Status err = (window_changed) ? ARM_COMPUTE_CREATE_ERROR(ErrorCode::RUNTIME_ERROR, "Insufficient Padding!") : Status{};
     return std::make_pair(err, win);
 }
 } // namespace
 
 CLGEMMLowpOffsetContributionKernel::CLGEMMLowpOffsetContributionKernel()
-    : _vector_sum_col(nullptr), _vector_sum_row(nullptr), _mm_result(nullptr)
+    : _vector_sum_col(nullptr), _vector_sum_row(nullptr), _mm_result(nullptr), _bias(nullptr)
 {
 }
 
-void CLGEMMLowpOffsetContributionKernel::configure(ICLTensor *mm_result, const ICLTensor *vector_sum_col, const ICLTensor *vector_sum_row, int32_t k, int32_t a_offset, int32_t b_offset)
+void CLGEMMLowpOffsetContributionKernel::configure(ICLTensor *mm_result, const ICLTensor *vector_sum_col, const ICLTensor *vector_sum_row, const ICLTensor *bias, int32_t k, int32_t a_offset,
+                                                   int32_t b_offset)
 {
     // Perform validate step
     ARM_COMPUTE_ERROR_ON_NULLPTR(mm_result);
     ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(mm_result->info(),
                                                   vector_sum_col != nullptr ? vector_sum_col->info() : nullptr,
                                                   vector_sum_row != nullptr ? vector_sum_row->info() : nullptr,
+                                                  bias != nullptr ? bias->info() : nullptr,
                                                   a_offset, b_offset)); // NOLINT
 
     _vector_sum_col = vector_sum_col;
     _vector_sum_row = vector_sum_row;
     _mm_result      = mm_result;
+    _bias           = bias;
 
     // Check if input is a 3D reinterpretation
     const bool reinterpret_as_3d = vector_sum_row != nullptr
@@ -161,20 +177,24 @@
     build_opts.add_option("-DK_OFFSET=" + support::cpp11::to_string(a_offset * b_offset * k));
     build_opts.add_option_if(reinterpret_as_3d, "-DHEIGHT_INPUT3D=" + support::cpp11::to_string(mm_result->info()->dimension(1)));
     build_opts.add_option_if(reinterpret_as_3d, "-DDEPTH_INPUT3D=" + support::cpp11::to_string(mm_result->info()->dimension(2)));
+    build_opts.add_option_if(bias != nullptr, "-DADD_BIAS");
+
+    std::string kernel_name("gemmlowp_offset_contribution");
 
     // Create kernel
-    _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel("gemmlowp_offset_contribution", build_opts.options()));
+    _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel(kernel_name, build_opts.options()));
 
     // Configure kernel window
     auto win_config = validate_and_configure_window(mm_result->info(),
                                                     vector_sum_col != nullptr ? vector_sum_col->info() : nullptr,
                                                     vector_sum_row != nullptr ? vector_sum_row->info() : nullptr,
+                                                    bias != nullptr ? bias->info() : nullptr,
                                                     a_offset, b_offset); // NOLINT
     ARM_COMPUTE_ERROR_THROW_ON(win_config.first);
     ICLKernel::configure_internal(win_config.second);
 
     // Set config_id for enabling LWS tuning
-    _config_id = "gemmlowp_offset_contribution_";
+    _config_id = kernel_name + "_";
     _config_id += support::cpp11::to_string(mm_result->info()->dimension(0));
     _config_id += "_";
     _config_id += support::cpp11::to_string(mm_result->info()->dimension(1));
@@ -182,13 +202,14 @@
     _config_id += support::cpp11::to_string(mm_result->info()->dimension(2));
 }
 
-Status CLGEMMLowpOffsetContributionKernel::validate(const ITensorInfo *mm_result, const ITensorInfo *vector_sum_col, const ITensorInfo *vector_sum_row,
+Status CLGEMMLowpOffsetContributionKernel::validate(const ITensorInfo *mm_result, const ITensorInfo *vector_sum_col, const ITensorInfo *vector_sum_row, const ITensorInfo *bias,
                                                     int32_t a_offset, int32_t b_offset)
 {
-    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(mm_result, vector_sum_col, vector_sum_row, a_offset, b_offset));
+    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(mm_result, vector_sum_col, vector_sum_row, bias, a_offset, b_offset));
     ARM_COMPUTE_RETURN_ON_ERROR(validate_and_configure_window(mm_result->clone().get(),
                                                               vector_sum_col != nullptr ? vector_sum_col->clone().get() : nullptr,
                                                               vector_sum_row != nullptr ? vector_sum_row->clone().get() : nullptr,
+                                                              bias != nullptr ? bias->clone().get() : nullptr,
                                                               a_offset, b_offset)
                                 .first); // NOLINT
 
@@ -214,6 +235,10 @@
     win_vector_sum_row.set(Window::DimY, Window::Dimension(0, 0, 0));
     win_vector_sum_col.set(Window::DimZ, Window::Dimension(0, 0, 0));
 
+    Window biases_slice = slice;
+    biases_slice.set(Window::DimY, Window::Dimension(0, 1, 1));
+    biases_slice.set(Window::DimZ, Window::Dimension(0, 1, 1));
+
     do
     {
         unsigned int idx = 0;
@@ -226,7 +251,11 @@
         {
             add_2D_tensor_argument(idx, _vector_sum_row, win_vector_sum_row);
         }
-        enqueue(queue, *this, slice);
+        if(_bias != nullptr)
+        {
+            add_1D_tensor_argument(idx, _bias, biases_slice);
+        }
+        enqueue(queue, *this, slice, lws_hint());
     }
     while(collapsed.slide_window_slice_3D(slice));
 }