diff --git a/compiler/luci-interpreter/pal/linux/KernelsToBuild.lst b/compiler/luci-interpreter/pal/linux/KernelsToBuild.lst
index 4f2ea2f2a43..df47427a3df 100644
--- a/compiler/luci-interpreter/pal/linux/KernelsToBuild.lst
+++ b/compiler/luci-interpreter/pal/linux/KernelsToBuild.lst
@@ -67,6 +67,7 @@ REGISTER_KERNEL(ResizeBilinear)
 REGISTER_KERNEL(ResizeNearestNeighbor)
 REGISTER_KERNEL(ReverseV2)
 REGISTER_KERNEL(RmsNorm)
+REGISTER_KERNEL(RoPE)
 REGISTER_KERNEL(Rsqrt)
 REGISTER_KERNEL(Select)
 REGISTER_KERNEL(SelectV2)
diff --git a/compiler/luci-interpreter/src/core/KernelParams.h b/compiler/luci-interpreter/src/core/KernelParams.h
index 100a09b7544..cc6a83e08c7 100644
--- a/compiler/luci-interpreter/src/core/KernelParams.h
+++ b/compiler/luci-interpreter/src/core/KernelParams.h
@@ -20,6 +20,7 @@
 #include <luci/IR/AttrPadding.h>
 #include <luci/IR/AttrFusedActFunc.h>
 #include <luci/IR/AttrMirrorPadMode.h>
+#include <luci/IR/AttrRoPEMode.h>
 #include <luci_interpreter/core/DataType.h>
 
 #include <cstdint>
@@ -32,6 +33,7 @@ namespace luci_interpreter
 using Activation = luci::FusedActFunc;
 using Padding = luci::Padding;
 using MirrorPadMode = luci::MirrorPadMode;
+using RoPEMode = luci::RoPEMode;
 
 struct AddParams
 {
@@ -186,6 +188,11 @@ struct RmsNormParams
   float epsilon;
 };
 
+struct RoPEParams
+{
+  RoPEMode mode;
+};
+
 struct ShapeParams
 {
   loco::DataType out_type;
diff --git a/compiler/luci-interpreter/src/kernels/RoPE.cpp b/compiler/luci-interpreter/src/kernels/RoPE.cpp
new file mode 100644
index 00000000000..f86ade6817d
--- /dev/null
+++ b/compiler/luci-interpreter/src/kernels/RoPE.cpp
@@ -0,0 +1,99 @@
+/*
+ * Copyright (c) 2024 Samsung Electronics Co., Ltd. All Rights Reserved
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "kernels/RoPE.h"
+
+#include "kernels/Utils.h"
+
+namespace luci_interpreter
+{
+namespace kernels
+{
+
+RoPE::RoPE(const Tensor *input, const Tensor *sin_table, const Tensor *cos_table, Tensor *output,
+           const RoPEParams &params)
+  : KernelWithParams<RoPEParams>({input, sin_table, cos_table}, {output}, params)
+{
+}
+
+void RoPE::configure()
+{
+  LUCI_INTERPRETER_CHECK(input()->shape().num_dims() == 4);
+  LUCI_INTERPRETER_CHECK(sin_table()->shape().dim(3) == input()->shape().dim(3));
+  LUCI_INTERPRETER_CHECK(cos_table()->shape().dim(3) == input()->shape().dim(3));
+
+  LUCI_INTERPRETER_CHECK(params().mode == RoPEMode::GPT_NEOX);
+
+  output()->resize(input()->shape());
+}
+
+void RoPE::execute() const
+{
+  switch (input()->element_type())
+  {
+    case DataType::FLOAT32:
+      evalFloat();
+      break;
+    default:
+      throw std::runtime_error("luci-rope Unsupported data type.");
+  }
+}
+
+void RoPE::evalFloat() const
+{
+  const auto input_shape = getTensorShape(input());
+  const auto sin_table_shape = getTensorShape(sin_table());
+  const auto cos_table_shape = getTensorShape(cos_table());
+  auto output_shape = getTensorShape(output());
+
+  const float *input_data = getTensorData<float>(input());
+  const float *sin_table_data = getTensorData<float>(sin_table());
+  const float *cos_table_data = getTensorData<float>(cos_table());
+  float *output_data = getTensorData<float>(output());
+
+  if (params().mode == RoPEMode::GPT_NEOX)
+  {
+    const int32_t i0_n = input_shape.Dims(0);
+    const int32_t i1_n = input_shape.Dims(1); // multihead
+    const int32_t i2_n = input_shape.Dims(2);
+    const int32_t i3_n = input_shape.Dims(3); // head
+
+    for (int32_t i0 = 0; i0 < i0_n; ++i0)
+    {
+      for (int32_t i1 = 0; i1 < i1_n; ++i1)
+      {
+        for (int32_t i2 = 0; i2 < i2_n; ++i2)
+        {
+          for (int32_t i3 = 0; i3 < i3_n / 2; ++i3)
+          {
+            const int32_t offset = tflite::Offset(input_shape, i0, i1, i2, i3);
+            const float x0 = input_data[offset];
+            const float x1 = input_data[offset + i3_n / 2];
+
+            output_data[offset] = x0 * cos_table_data[i3] - x1 * sin_table_data[i3];
+            output_data[offset + i3_n / 2] =
+              x0 * sin_table_data[i3 + i3_n / 2] + x1 * cos_table_data[i3 + i3_n / 2];
+          }
+        }
+      }
+    }
+  }
+  else
+    throw std::runtime_error("luci-intp RoPE unsupported mode.");
+}
+
+} // namespace kernels
+} // namespace luci_interpreter
diff --git a/compiler/luci-interpreter/src/kernels/RoPE.h b/compiler/luci-interpreter/src/kernels/RoPE.h
new file mode 100644
index 00000000000..5576ece5e11
--- /dev/null
+++ b/compiler/luci-interpreter/src/kernels/RoPE.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright (c) 2024 Samsung Electronics Co., Ltd. All Rights Reserved
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef LUCI_INTERPRETER_KERNELS_ROPE_H
+#define LUCI_INTERPRETER_KERNELS_ROPE_H
+
+#include "core/Kernel.h"
+#include "core/KernelParams.h"
+
+namespace luci_interpreter
+{
+namespace kernels
+{
+
+class RoPE : public KernelWithParams<RoPEParams>
+{
+public:
+  RoPE(const Tensor *input, const Tensor *sin_table, const Tensor *cos_table, Tensor *output,
+       const RoPEParams &params);
+
+  const Tensor *input() const { return _inputs[0]; }
+  const Tensor *sin_table() const { return _inputs[1]; }
+  const Tensor *cos_table() const { return _inputs[2]; }
+  Tensor *output() const { return _outputs[0]; }
+
+  void configure() override;
+  void execute() const override;
+
+private:
+  void evalFloat() const;
+};
+
+} // namespace kernels
+} // namespace luci_interpreter
+
+#endif // LUCI_INTERPRETER_KERNELS_ROPE_H
diff --git a/compiler/luci-interpreter/src/kernels/RoPE.test.cpp b/compiler/luci-interpreter/src/kernels/RoPE.test.cpp
new file mode 100644
index 00000000000..e3ab10abde7
--- /dev/null
+++ b/compiler/luci-interpreter/src/kernels/RoPE.test.cpp
@@ -0,0 +1,186 @@
+/*
+ * Copyright (c) 2024 Samsung Electronics Co., Ltd. All Rights Reserved
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "kernels/RoPE.h"
+#include "kernels/TestUtils.h"
+#include "luci_interpreter/TestMemoryManager.h"
+
+namespace luci_interpreter
+{
+namespace kernels
+{
+namespace
+{
+
+using namespace testing;
+
+class RoPETest : public ::testing::Test
+{
+protected:
+  void SetUp() override { _memory_manager = std::make_unique<TestMemoryManager>(); }
+
+  std::unique_ptr<IMemoryManager> _memory_manager;
+};
+
+TEST_F(RoPETest, floatTest)
+{
+  Shape input_shape{1, 1, 1, 4};
+  std::vector<float> input_data{0, 1.0, 2.0, 3.0};
+
+  Shape sin_shape{1, 1, 1, 4};
+  std::vector<float> sin_data{0.5, 1.0, 1.0, 0.5};
+
+  Shape cos_shape{1, 1, 1, 4};
+  std::vector<float> cos_data{1.0, 0.5, 0.5, 1.0};
+
+  Shape ref_output_shape{1, 1, 1, 4};
+  std::vector<float> ref_output_data{-1.0, -2.5, 1.0, 3.5};
+
+  Tensor input_tensor =
+    makeInputTensor<DataType::FLOAT32>(input_shape, input_data, _memory_manager.get());
+  Tensor output_tensor = makeOutputTensor(DataType::FLOAT32);
+
+  Tensor sin_table = makeInputTensor<DataType::FLOAT32>(sin_shape, sin_data, _memory_manager.get());
+  Tensor cos_table = makeInputTensor<DataType::FLOAT32>(cos_shape, cos_data, _memory_manager.get());
+
+  RoPEParams params{};
+  params.mode = RoPEMode::GPT_NEOX;
+
+  RoPE kernel(&input_tensor, &sin_table, &cos_table, &output_tensor, params);
+  kernel.configure();
+  _memory_manager->allocate_memory(output_tensor);
+  kernel.execute();
+
+  EXPECT_THAT(extractTensorData<float>(output_tensor),
+              ::testing::ElementsAreArray(ref_output_data));
+  EXPECT_THAT(extractTensorShape(output_tensor), ::testing::ElementsAreArray({1, 1, 1, 4}));
+}
+
+TEST_F(RoPETest, Unsupported_dims_NEG)
+{
+  Shape input_shape{1, 1, 3};
+  std::vector<float> input_data{0, 1.0, 2.0};
+
+  Shape sin_shape{1, 1, 3};
+  std::vector<float> sin_data{0.5, 1.0, 1.0};
+
+  Shape cos_shape{1, 1, 3};
+  std::vector<float> cos_data{1.0, 0.5, 0.5};
+
+  Shape ref_output_shape{1, 1, 3};
+  std::vector<float> ref_output_data{-1.0, -2.5, 1.0};
+
+  Tensor input_tensor =
+    makeInputTensor<DataType::FLOAT32>(input_shape, input_data, _memory_manager.get());
+  Tensor output_tensor = makeOutputTensor(DataType::FLOAT32);
+
+  Tensor sin_table = makeInputTensor<DataType::FLOAT32>(sin_shape, sin_data, _memory_manager.get());
+  Tensor cos_table = makeInputTensor<DataType::FLOAT32>(cos_shape, cos_data, _memory_manager.get());
+
+  RoPEParams params{};
+  params.mode = RoPEMode::GPT_NEOX;
+
+  RoPE kernel(&input_tensor, &sin_table, &cos_table, &output_tensor, params);
+  EXPECT_ANY_THROW(kernel.configure());
+}
+
+TEST_F(RoPETest, Unsupported_mode_NEG)
+{
+  Shape input_shape{1, 1, 1, 4};
+  std::vector<float> input_data{0, 1.0, 2.0, 3.0};
+
+  Shape sin_shape{1, 1, 1, 4};
+  std::vector<float> sin_data{0.5, 1.0, 1.0, 0.5};
+
+  Shape cos_shape{1, 1, 1, 4};
+  std::vector<float> cos_data{1.0, 0.5, 0.5, 1.0};
+
+  Shape ref_output_shape{1, 1, 1, 4};
+  std::vector<float> ref_output_data{-1.0, -2.5, 1.0, 3.5};
+
+  Tensor input_tensor =
+    makeInputTensor<DataType::FLOAT32>(input_shape, input_data, _memory_manager.get());
+  Tensor output_tensor = makeOutputTensor(DataType::FLOAT32);
+
+  Tensor sin_table = makeInputTensor<DataType::FLOAT32>(sin_shape, sin_data, _memory_manager.get());
+  Tensor cos_table = makeInputTensor<DataType::FLOAT32>(cos_shape, cos_data, _memory_manager.get());
+
+  RoPEParams params{};
+  params.mode = RoPEMode::GPT_J;
+
+  RoPE kernel(&input_tensor, &sin_table, &cos_table, &output_tensor, params);
+  EXPECT_ANY_THROW(kernel.configure());
+}
+
+TEST_F(RoPETest, Invalid_input_sin_table_NEG)
+{
+  Shape input_shape{1, 1, 1, 4};
+  std::vector<float> input_data{0, 1.0, 2.0, 3.0};
+
+  Shape sin_shape{1, 1, 1, 3};
+  std::vector<float> sin_data{0.5, 1.0, 1.0};
+
+  Shape cos_shape{1, 1, 1, 4};
+  std::vector<float> cos_data{1.0, 0.5, 0.5, 1.0};
+
+  Shape ref_output_shape{1, 1, 1, 4};
+  std::vector<float> ref_output_data{-1.0, -2.5, 1.0, 3.5};
+
+  Tensor input_tensor =
+    makeInputTensor<DataType::FLOAT32>(input_shape, input_data, _memory_manager.get());
+  Tensor output_tensor = makeOutputTensor(DataType::FLOAT32);
+
+  Tensor sin_table = makeInputTensor<DataType::FLOAT32>(sin_shape, sin_data, _memory_manager.get());
+  Tensor cos_table = makeInputTensor<DataType::FLOAT32>(cos_shape, cos_data, _memory_manager.get());
+
+  RoPEParams params{};
+  params.mode = RoPEMode::GPT_NEOX;
+
+  RoPE kernel(&input_tensor, &sin_table, &cos_table, &output_tensor, params);
+  EXPECT_ANY_THROW(kernel.configure());
+}
+
+TEST_F(RoPETest, Invalid_input_cos_table_NEG)
+{
+  Shape input_shape{1, 1, 1, 4};
+  std::vector<float> input_data{0, 1.0, 2.0, 3.0};
+
+  Shape sin_shape{1, 1, 1, 4};
+  std::vector<float> sin_data{0.5, 1.0, 1.0, 0.5};
+
+  Shape cos_shape{1, 1, 1, 3};
+  std::vector<float> cos_data{1.0, 0.5, 0.5};
+
+  Shape ref_output_shape{1, 1, 1, 4};
+  std::vector<float> ref_output_data{-1.0, -2.5, 1.0, 3.5};
+
+  Tensor input_tensor =
+    makeInputTensor<DataType::FLOAT32>(input_shape, input_data, _memory_manager.get());
+  Tensor output_tensor = makeOutputTensor(DataType::FLOAT32);
+
+  Tensor sin_table = makeInputTensor<DataType::FLOAT32>(sin_shape, sin_data, _memory_manager.get());
+  Tensor cos_table = makeInputTensor<DataType::FLOAT32>(cos_shape, cos_data, _memory_manager.get());
+
+  RoPEParams params{};
+  params.mode = RoPEMode::GPT_NEOX;
+
+  RoPE kernel(&input_tensor, &sin_table, &cos_table, &output_tensor, params);
+  EXPECT_ANY_THROW(kernel.configure());
+}
+
+} // namespace
+} // namespace kernels
+} // namespace luci_interpreter
diff --git a/compiler/luci-interpreter/src/loader/KernelBuilder.test.cpp b/compiler/luci-interpreter/src/loader/KernelBuilder.test.cpp
index 3e05da81e45..0a3c30fff1f 100644
--- a/compiler/luci-interpreter/src/loader/KernelBuilder.test.cpp
+++ b/compiler/luci-interpreter/src/loader/KernelBuilder.test.cpp
@@ -69,6 +69,7 @@
 #include <kernels/ResizeNearestNeighbor.h>
 #include <kernels/ReverseV2.h>
 #include <kernels/RmsNorm.h>
+#include <kernels/RoPE.h>
 #include <kernels/Rsqrt.h>
 #include <kernels/Sin.h>
 #include <kernels/Slice.h>
@@ -1118,6 +1119,26 @@ TEST_F(KernelBuilderTest, RmsNorm)
   EXPECT_THAT(kernel->params().epsilon, Eq(op->epsilon()));
 }
 
+TEST_F(KernelBuilderTest, RoPE)
+{
+  auto *input = createInputNode();
+  auto *sin_table = createInputNode();
+  auto *cos_table = createInputNode();
+
+  auto *op = createNode<luci::CircleRoPE>();
+  op->input(input);
+  op->sin_table(sin_table);
+  op->cos_table(cos_table);
+
+  auto kernel = buildKernel<kernels::RoPE>(op);
+  ASSERT_THAT(kernel, NotNull());
+
+  checkTensor(kernel->input(), input);
+  checkTensor(kernel->sin_table(), sin_table);
+  checkTensor(kernel->cos_table(), cos_table);
+  checkTensor(kernel->output(), op);
+}
+
 TEST_F(KernelBuilderTest, Rsqrt)
 {
   auto *input = createInputNode();
diff --git a/compiler/luci-interpreter/src/loader/nodes/RoPE.cpp b/compiler/luci-interpreter/src/loader/nodes/RoPE.cpp
new file mode 100644
index 00000000000..722015d6d2d
--- /dev/null
+++ b/compiler/luci-interpreter/src/loader/nodes/RoPE.cpp
@@ -0,0 +1,42 @@
+/*
+ * Copyright (c) 2024 Samsung Electronics Co., Ltd. All Rights Reserved
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *    http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include "Builders.h"
+
+#include "kernels/RoPE.h"
+
+namespace luci_interpreter
+{
+
+std::unique_ptr<Kernel> build_kernel_CircleRoPE(const luci::CircleNode *circle_node,
+                                                KernelBuilderHelper &helper)
+{
+  const auto *node = loco::must_cast<const luci::CircleRoPE *>(circle_node);
+  assert(node->arity() == 3);
+
+  const Tensor *input = helper.getInputTensor(node->input());
+  const Tensor *sin_table = helper.getInputTensor(node->sin_table());
+  const Tensor *cos_table = helper.getInputTensor(node->cos_table());
+
+  Tensor *output = helper.getOutputTensor(node);
+
+  RoPEParams params{};
+  params.mode = node->mode();
+
+  return std::make_unique<kernels::RoPE>(input, sin_table, cos_table, output, params);
+}
+
+} // namespace luci_interpreter