diff --git a/src/BUILD b/src/BUILD
index 80088630ff..2e7a295cc5 100644
--- a/src/BUILD
+++ b/src/BUILD
@@ -602,10 +602,6 @@ cc_library(
         deps = select({
             "//:not_disable_python": [
                 "//src/python:libovmspythonmodule",
-                # Jinja template processing is done in Python
-                "//src/llm:llmcalculator",
-                "//src/llm:genai_servables",
-                "//src/llm:text_processor",
             ],
             "//:disable_python": []
         }) + select({
@@ -613,7 +609,8 @@ cc_library(
             "//:not_disable_mediapipe" : [
                 "//src/llm:openai_completions_api_handler",
                 "//src/embeddings:embeddingscalculator",
-                "//src/rerank:rerankcalculator",],
+                "//src/rerank:rerankcalculator",
+                "//src/llm:llmcalculator",],
         }) + select({
             "//:enable_drogon": ["libdrogon_http_server"],
             "//conditions:default" : ["libnet_http_server"],
@@ -2756,6 +2753,11 @@ cc_test(
                 "test/get_mediapipe_graph_metadata_response_test.cpp",
                 "test/mediapipe_framework_test.cpp",
                 "test/http_openai_handler_test.cpp",
+                "test/llm/llmnode_test.cpp",
+                "test/llm/max_model_length_test.cpp",
+                "test/llm/text_streamer_test.cpp",
+                "test/llm/visual_language_model/complete_flow_test.cpp",
+                "test/llm/visual_language_model/initialization_test.cpp",
             ],
             "//:disable_mediapipe" : [
                 "test/disabled_mediapipe_test.cpp",
@@ -2765,13 +2767,8 @@ cc_test(
                 # OvmsPyTensor is currently not used in OVMS core and is just a base for the binding.
                 # "test/python/ovms_py_tensor_test.cpp",
                 "test/pythonnode_test.cpp",
-                # LLM logic uses Python for processing Jinja templates
-                "test/llm/llmnode_test.cpp",
-                "test/llm/max_model_length_test.cpp",
+                # LLM logic uses Python for processing Jinja templates when built with Python enabled
                 "test/llm/llmtemplate_test.cpp",
-                "test/llm/text_streamer_test.cpp",
-                "test/llm/visual_language_model/complete_flow_test.cpp",
-                "test/llm/visual_language_model/initialization_test.cpp",
             ],
             "//:disable_python" : [],
         }),
diff --git a/src/llm/BUILD b/src/llm/BUILD
index 3f3ce60ed9..030655350e 100644
--- a/src/llm/BUILD
+++ b/src/llm/BUILD
@@ -16,9 +16,9 @@
 
 load("@mediapipe//mediapipe/framework/port:build_config.bzl", "mediapipe_cc_proto_library", "mediapipe_proto_library")
 load("//:common_settings.bzl",
-     "COMMON_STATIC_LIBS_COPTS", "COMMON_STATIC_LIBS_LINKOPTS", "COMMON_FUZZER_COPTS", "COMMON_FUZZER_LINKOPTS", "COMMON_LOCAL_DEFINES", "PYBIND_DEPS")
+     "COMMON_STATIC_LIBS_COPTS", "COMMON_STATIC_LIBS_LINKOPTS", "COMMON_FUZZER_COPTS", "COMMON_FUZZER_LINKOPTS", "COMMON_LOCAL_DEFINES", "PYBIND_DEPS", "COPTS_PYTHON")
 
-COPTS_ADJUSTED = COMMON_STATIC_LIBS_COPTS + select({
+COPTS_ADJUSTED = COMMON_STATIC_LIBS_COPTS + COPTS_PYTHON + select({
         "//conditions:default": [],
         "//:fuzzer_build" : COMMON_FUZZER_COPTS,
 })
@@ -92,13 +92,30 @@ cc_library(
 
 cc_library(
     name = "genai_servables",
-    hdrs = ["servable.hpp", "servable_initializer.hpp", 
-            "language_model/continuous_batching/servable.hpp", "language_model/continuous_batching/llm_executor.hpp", "language_model/continuous_batching/servable_initializer.hpp",
-            "visual_language_model/continuous_batching/servable.hpp", "language_model/legacy/servable.hpp", "language_model/legacy/servable_initializer.hpp", "language_model/legacy/legacy_executor.hpp",
-            "visual_language_model/legacy/servable.hpp", "visual_language_model/legacy/servable_initializer.hpp", "visual_language_model/legacy/legacy_executor.hpp"],
-    srcs = ["servable.cpp", "servable_initializer.cpp", "language_model/continuous_batching/servable.cpp", "language_model/continuous_batching/servable_initializer.cpp",
-            "visual_language_model/continuous_batching/servable.cpp", "language_model/legacy/servable.cpp", "language_model/legacy/servable_initializer.cpp", "language_model/legacy/legacy_executor.cpp",
-            "visual_language_model/legacy/servable.cpp", "visual_language_model/legacy/servable_initializer.cpp", "visual_language_model/legacy/legacy_executor.cpp"],
+    hdrs = ["servable.hpp", 
+            "servable_initializer.hpp", 
+            "language_model/continuous_batching/servable.hpp",
+            "language_model/continuous_batching/llm_executor.hpp",
+            "language_model/continuous_batching/servable_initializer.hpp",
+            "visual_language_model/continuous_batching/servable.hpp",
+            "language_model/legacy/servable.hpp",
+            "language_model/legacy/servable_initializer.hpp",
+            "language_model/legacy/legacy_executor.hpp",
+            "visual_language_model/legacy/servable.hpp",
+            "visual_language_model/legacy/servable_initializer.hpp",
+            "visual_language_model/legacy/legacy_executor.hpp",
+            "text_utils.hpp"],
+    srcs = ["servable.cpp",
+            "servable_initializer.cpp",
+            "language_model/continuous_batching/servable.cpp",
+            "language_model/continuous_batching/servable_initializer.cpp",
+            "visual_language_model/continuous_batching/servable.cpp",
+            "language_model/legacy/servable.cpp",
+            "language_model/legacy/servable_initializer.cpp",
+            "language_model/legacy/legacy_executor.cpp",
+            "visual_language_model/legacy/servable.cpp",
+            "visual_language_model/legacy/servable_initializer.cpp",
+            "visual_language_model/legacy/legacy_executor.cpp"],
     deps = [
         "//third_party:openvino",
         "@mediapipe//mediapipe/framework:calculator_framework",
@@ -110,14 +127,15 @@ cc_library(
         "//src:libovmsprofiler",
         "//src:libovmsfilesystem",
         "llmcalculator_cc_proto",
-        "//src/python:utils",
-        ":text_processor",
         ":openai_completions_api_handler",
         "//src:httppayload",
         "//src:libhttpclientconnection",
-    ] + PYBIND_DEPS + select({
+    ] + select({
         "//conditions:default": ["//third_party:genai", ":llm_engine"],
         "//:not_genai_bin" : [":llm_engine"],
+    }) + select({
+        "//:disable_python": [],
+        "//:not_disable_python" : [":py_jinja_template_processor"],
     }),
     visibility = ["//visibility:public"],
     local_defines = COMMON_LOCAL_DEFINES,
@@ -127,9 +145,9 @@ cc_library(
 )
 
 cc_library(
-    name = "text_processor",
-    hdrs = ["text_processor.hpp"],
-    srcs = ["text_processor.cpp"],
+    name = "py_jinja_template_processor",
+    hdrs = ["py_jinja_template_processor.hpp"],
+    srcs = ["py_jinja_template_processor.cpp"],
     deps = ["@mediapipe//mediapipe/framework:calculator_framework",
             "//third_party:openvino",
             "//src:libovmslogging",
diff --git a/src/llm/http_llm_calculator.cc b/src/llm/http_llm_calculator.cc
index d1ee639762..ba9260bcab 100644
--- a/src/llm/http_llm_calculator.cc
+++ b/src/llm/http_llm_calculator.cc
@@ -26,6 +26,7 @@
 #pragma warning(pop)
 
 #include "../http_payload.hpp"
+#include "../logging.hpp"
 #include "../profiler.hpp"
 #include "apis/openai_completions.hpp"
 #include "servable.hpp"
diff --git a/src/llm/language_model/continuous_batching/servable.cpp b/src/llm/language_model/continuous_batching/servable.cpp
index fa62fb8d5b..3ab4850bb8 100644
--- a/src/llm/language_model/continuous_batching/servable.cpp
+++ b/src/llm/language_model/continuous_batching/servable.cpp
@@ -33,7 +33,10 @@
 #include "../../../http_payload.hpp"
 #include "../../../mediapipe_internal/mediapipe_utils.hpp"
 #include "../../apis/openai_completions.hpp"
-#include "../../text_processor.hpp"
+#include "../../text_utils.hpp"
+#if (PYTHON_DISABLE == 0)
+#include "../../py_jinja_template_processor.hpp"
+#endif
 #include "llm_executor.hpp"
 #include "servable.hpp"
 
diff --git a/src/llm/language_model/continuous_batching/servable_initializer.cpp b/src/llm/language_model/continuous_batching/servable_initializer.cpp
index ff608b7ac9..905bc16b05 100644
--- a/src/llm/language_model/continuous_batching/servable_initializer.cpp
+++ b/src/llm/language_model/continuous_batching/servable_initializer.cpp
@@ -115,8 +115,9 @@ Status ContinuousBatchingServableInitializer::initializeExperimental(std::shared
         SPDLOG_ERROR("Error during llm node initialization for models_path: {}", parsedModelsPath);
         return StatusCode::LLM_NODE_RESOURCE_STATE_INITIALIZATION_FAILED;
     }
-
-    loadTextProcessor(properties, parsedModelsPath);
+#if (PYTHON_DISABLE == 0)
+    loadTemplateProcessor(properties, parsedModelsPath);
+#endif
     if (nodeOptions.has_max_tokens_limit()) {
         properties->maxTokensLimit = nodeOptions.max_tokens_limit();
     }
@@ -133,15 +134,12 @@ Status ContinuousBatchingServableInitializer::initialize(std::shared_ptr<GenAiSe
         return status;
     }
     auto properties = std::static_pointer_cast<ContinuousBatchingServableProperties>(servable->getProperties());
-
     properties->modelsPath = parsedModelsPath;
-
     properties->schedulerConfig.max_num_batched_tokens = nodeOptions.max_num_batched_tokens();
     properties->schedulerConfig.cache_size = nodeOptions.cache_size();
     properties->schedulerConfig.dynamic_split_fuse = nodeOptions.dynamic_split_fuse();
     properties->schedulerConfig.max_num_seqs = nodeOptions.max_num_seqs();
     properties->schedulerConfig.enable_prefix_caching = nodeOptions.enable_prefix_caching();
-
     properties->device = nodeOptions.device();
     properties->isSpeculativePipeline = false;
 
@@ -163,7 +161,6 @@ Status ContinuousBatchingServableInitializer::initialize(std::shared_ptr<GenAiSe
         properties->pluginConfig.insert(draftPipeline);
         properties->isSpeculativePipeline = true;
     } else if (nodeOptions.has_draft_max_num_batched_tokens() || nodeOptions.has_draft_cache_size() || nodeOptions.has_draft_dynamic_split_fuse() || nodeOptions.has_draft_max_num_seqs() || nodeOptions.has_draft_block_size() || nodeOptions.has_draft_device()) {
-        // Consider moving draft parameters to separate structure in node options, so it's validated on the proto level
         SPDLOG_ERROR("Draft model path is not provided, but draft scheduler options are set.");
         return StatusCode::LLM_NODE_RESOURCE_STATE_INITIALIZATION_FAILED;
     }
@@ -188,14 +185,16 @@ Status ContinuousBatchingServableInitializer::initialize(std::shared_ptr<GenAiSe
         return StatusCode::LLM_NODE_RESOURCE_STATE_INITIALIZATION_FAILED;
     }
 
-    loadTextProcessor(properties, parsedModelsPath);
+#if (PYTHON_DISABLE == 0)
+    loadTemplateProcessor(properties, parsedModelsPath);
+#endif
     if (nodeOptions.has_max_tokens_limit()) {
         properties->maxTokensLimit = nodeOptions.max_tokens_limit();
     }
     properties->bestOfLimit = nodeOptions.best_of_limit();
     properties->maxModelLength = parseMaxModelLength(parsedModelsPath);
-
     properties->llmExecutorWrapper = std::make_shared<LLMExecutorWrapper>(properties->pipeline);
+
     return StatusCode::OK;
 }
 
diff --git a/src/llm/language_model/legacy/servable.cpp b/src/llm/language_model/legacy/servable.cpp
index 7c3a76acaa..b09f00362d 100644
--- a/src/llm/language_model/legacy/servable.cpp
+++ b/src/llm/language_model/legacy/servable.cpp
@@ -33,7 +33,10 @@
 #include "../../../http_payload.hpp"
 #include "../../../mediapipe_internal/mediapipe_utils.hpp"
 #include "../../apis/openai_completions.hpp"
-#include "../../text_processor.hpp"
+#include "../../text_utils.hpp"
+#if (PYTHON_DISABLE == 0)
+#include "../../py_jinja_template_processor.hpp"
+#endif
 #include "servable.hpp"
 
 namespace ovms {
diff --git a/src/llm/language_model/legacy/servable_initializer.cpp b/src/llm/language_model/legacy/servable_initializer.cpp
index 10b61d9e39..e28f95bd57 100644
--- a/src/llm/language_model/legacy/servable_initializer.cpp
+++ b/src/llm/language_model/legacy/servable_initializer.cpp
@@ -96,8 +96,9 @@ Status LegacyServableInitializer::initialize(std::shared_ptr<GenAiServable>& ser
         SPDLOG_ERROR("Error during llm node initialization for models_path: {}", parsedModelsPath);
         return StatusCode::LLM_NODE_RESOURCE_STATE_INITIALIZATION_FAILED;
     }
-
-    loadTextProcessor(properties, parsedModelsPath);
+#if (PYTHON_DISABLE == 0)
+    loadTemplateProcessor(properties, parsedModelsPath);
+#endif
     properties->legacyExecutor = std::make_shared<LegacyExecutorWrapper>(properties->pipeline);
     if (nodeOptions.has_max_tokens_limit()) {
         properties->maxTokensLimit = nodeOptions.max_tokens_limit();
diff --git a/src/llm/text_processor.cpp b/src/llm/py_jinja_template_processor.cpp
similarity index 86%
rename from src/llm/text_processor.cpp
rename to src/llm/py_jinja_template_processor.cpp
index ce13fbcdfb..e4e717d074 100644
--- a/src/llm/text_processor.cpp
+++ b/src/llm/py_jinja_template_processor.cpp
@@ -13,7 +13,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 //*****************************************************************************
-#include "text_processor.hpp"
+#include "py_jinja_template_processor.hpp"
 
 #include <string>
 #include <utility>
@@ -35,16 +35,16 @@
 
 namespace ovms {
 
-bool TextProcessor::applyChatTemplate(TextProcessor& textProcessor, std::string modelsPath, const std::string& requestBody, std::string& output) {
-    if (textProcessor.chatTemplate == nullptr) {
+bool PyJinjaTemplateProcessor::applyChatTemplate(PyJinjaTemplateProcessor& templateProcessor, std::string modelsPath, const std::string& requestBody, std::string& output) {
+    if (templateProcessor.chatTemplate == nullptr) {
         output = "Error: Chat template not loaded correctly, so it cannot be applied";
         return false;
     }
 
     py::gil_scoped_acquire acquire;
     try {
-        auto locals = py::dict("request_body"_a = requestBody, "chat_template"_a = textProcessor.chatTemplate->getObject(),
-            "bos_token"_a = textProcessor.bosToken, "eos_token"_a = textProcessor.eosToken);
+        auto locals = py::dict("request_body"_a = requestBody, "chat_template"_a = templateProcessor.chatTemplate->getObject(),
+            "bos_token"_a = templateProcessor.bosToken, "eos_token"_a = templateProcessor.eosToken);
         py::exec(R"(
             output = ""
             error = ""
diff --git a/src/llm/py_jinja_template_processor.hpp b/src/llm/py_jinja_template_processor.hpp
new file mode 100644
index 0000000000..4837b69541
--- /dev/null
+++ b/src/llm/py_jinja_template_processor.hpp
@@ -0,0 +1,41 @@
+//*****************************************************************************
+// Copyright 2024 Intel Corporation
+//
+// 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.
+//*****************************************************************************
+#pragma once
+#include <memory>
+#include <sstream>
+#include <string>
+
+#include <openvino/openvino.hpp>
+#pragma warning(push)
+#pragma warning(disable : 6326 28182 6011 28020)
+// Python execution for template processing
+#include <pybind11/embed.h>  // everything needed for embedding
+#include <pybind11/stl.h>
+#pragma warning(pop)
+
+#include "src/python/utils.hpp"
+
+namespace ovms {
+
+class PyJinjaTemplateProcessor {
+public:
+    std::string bosToken = "";
+    std::string eosToken = "";
+    std::unique_ptr<PyObjectWrapper<py::object>> chatTemplate = nullptr;
+
+    static bool applyChatTemplate(PyJinjaTemplateProcessor& templateProcessor, std::string modelsPath, const std::string& requestBody, std::string& output);
+};
+}  // namespace ovms
diff --git a/src/llm/servable.cpp b/src/llm/servable.cpp
index 3818b12cfa..821c937d3c 100644
--- a/src/llm/servable.cpp
+++ b/src/llm/servable.cpp
@@ -32,7 +32,7 @@
 #include "../profiler.hpp"
 #include "apis/openai_completions.hpp"
 #include "servable.hpp"
-#include "text_processor.hpp"
+#include "text_utils.hpp"
 
 namespace ovms {
 absl::Status GenAiServable::loadRequest(std::shared_ptr<GenAiServableExecutionContext>& executionContext, const ovms::HttpPayload& payload) {
@@ -87,15 +87,21 @@ absl::Status GenAiServable::prepareInputs(std::shared_ptr<GenAiServableExecution
     std::string inputText;
     switch (executionContext->endpoint) {
     case Endpoint::CHAT_COMPLETIONS: {
+#if (PYTHON_DISABLE == 0)
         bool success;
         if (executionContext->apiHandler->getProcessedJson().size() > 0) {
-            success = TextProcessor::applyChatTemplate(getProperties()->textProcessor, getProperties()->modelsPath, executionContext->apiHandler->getProcessedJson(), inputText);
+            success = PyJinjaTemplateProcessor::applyChatTemplate(getProperties()->templateProcessor, getProperties()->modelsPath, executionContext->apiHandler->getProcessedJson(), inputText);
         } else {
-            success = TextProcessor::applyChatTemplate(getProperties()->textProcessor, getProperties()->modelsPath, executionContext->payload.body, inputText);
+            success = PyJinjaTemplateProcessor::applyChatTemplate(getProperties()->templateProcessor, getProperties()->modelsPath, executionContext->payload.body, inputText);
         }
         if (!success) {
             return absl::Status(absl::StatusCode::kInvalidArgument, inputText);
         }
+#else
+        ov::genai::ChatHistory& chatHistory = executionContext->apiHandler->getChatHistory();
+        constexpr bool add_generation_prompt = true;  // confirm it should be hardcoded
+        inputText = getProperties()->tokenizer.apply_chat_template(chatHistory, add_generation_prompt);
+#endif
         if (inputText.size() == 0) {
             return absl::Status(absl::StatusCode::kInvalidArgument, "Final prompt after applying chat template is empty");
         }
diff --git a/src/llm/servable.hpp b/src/llm/servable.hpp
index da3cd1fdab..03eb66c41c 100644
--- a/src/llm/servable.hpp
+++ b/src/llm/servable.hpp
@@ -32,7 +32,9 @@
 
 #include "../http_payload.hpp"
 #include "apis/openai_completions.hpp"
-#include "text_processor.hpp"
+#if (PYTHON_DISABLE == 0)
+#include "py_jinja_template_processor.hpp"
+#endif
 
 namespace ovms {
 // Some pipelines internals rely on request_id, so for now we provide increasing ID
@@ -81,12 +83,14 @@ struct GenAiServableProperties {
     ov::AnyMap tokenizerPluginConfig;
     // Sampling limits
     std::optional<uint32_t> maxTokensLimit;
+    std::optional<uint32_t> maxModelLength;
     uint32_t bestOfLimit;
     bool isSpeculativePipeline;  // sampling is generally common, but maybe we could avoid having this field at all
     // Text processing utilities
     ov::genai::Tokenizer tokenizer;
-    TextProcessor textProcessor;
-    std::optional<uint32_t> maxModelLength;
+#if (PYTHON_DISABLE == 0)
+    PyJinjaTemplateProcessor templateProcessor;
+#endif
 };
 
 class GenAiServable {
diff --git a/src/llm/servable_initializer.cpp b/src/llm/servable_initializer.cpp
index 96bb6317ea..2ae9c339d5 100644
--- a/src/llm/servable_initializer.cpp
+++ b/src/llm/servable_initializer.cpp
@@ -48,9 +48,10 @@
 
 namespace ovms {
 
+#if (PYTHON_DISABLE == 0)
 static const std::string CHAT_TEMPLATE_WARNING_MESSAGE = "Warning: Chat template has not been loaded properly. Servable will not respond to /chat/completions endpoint.";
 
-void GenAiServableInitializer::loadTextProcessor(std::shared_ptr<GenAiServableProperties> properties, const std::string& chatTemplateDirectory) {
+void GenAiServableInitializer::loadTemplateProcessor(std::shared_ptr<GenAiServableProperties> properties, const std::string& chatTemplateDirectory) {
     py::gil_scoped_acquire acquire;
     try {
         auto locals = py::dict("templates_directory"_a = chatTemplateDirectory);
@@ -108,9 +109,9 @@ void GenAiServableInitializer::loadTextProcessor(std::shared_ptr<GenAiServablePr
         )",
             py::globals(), locals);
 
-        properties->textProcessor.bosToken = locals["bos_token"].cast<std::string>();
-        properties->textProcessor.eosToken = locals["eos_token"].cast<std::string>();
-        properties->textProcessor.chatTemplate = std::make_unique<PyObjectWrapper<py::object>>(locals["template"]);
+        properties->templateProcessor.bosToken = locals["bos_token"].cast<std::string>();
+        properties->templateProcessor.eosToken = locals["eos_token"].cast<std::string>();
+        properties->templateProcessor.chatTemplate = std::make_unique<PyObjectWrapper<py::object>>(locals["template"]);
     } catch (const pybind11::error_already_set& e) {
         SPDLOG_INFO(CHAT_TEMPLATE_WARNING_MESSAGE);
         SPDLOG_DEBUG("Chat template loading failed with error: {}", e.what());
@@ -125,6 +126,7 @@ void GenAiServableInitializer::loadTextProcessor(std::shared_ptr<GenAiServablePr
         SPDLOG_DEBUG("Chat template loading failed with an unexpected error");
     }
 }
+#endif
 
 Status parseModelsPath(std::string& outPath, std::string modelsPath, std::string graphPath) {
     auto fsModelsPath = std::filesystem::path(modelsPath);
@@ -150,6 +152,31 @@ Status parseModelsPath(std::string& outPath, std::string modelsPath, std::string
     return StatusCode::OK;
 }
 
+std::optional<uint32_t> parseMaxModelLength(std::string& modelsPath) {
+    std::string configPath = FileSystem::appendSlash(modelsPath) + "config.json";
+    std::optional<uint32_t> maxModelLength;
+    if (std::filesystem::exists(configPath.c_str())) {
+        std::ifstream ifs(configPath);
+        if (!ifs.is_open()) {
+            return maxModelLength;
+        }
+        rapidjson::Document modelConfig;
+        rapidjson::IStreamWrapper isw(ifs);
+        rapidjson::ParseResult parseResult = modelConfig.ParseStream(isw);
+        if (parseResult.Code()) {
+            return maxModelLength;
+        }
+        std::vector<std::string> maxLengthFields = {"max_position_embeddings", "n_positions", "seq_len", "seq_length", "n_ctx", "sliding_window"};
+        for (auto field : maxLengthFields) {
+            if (modelConfig.HasMember(field.c_str()) && modelConfig[field.c_str()].IsUint()) {
+                maxModelLength = modelConfig[field.c_str()].GetUint();
+                break;
+            }
+        }
+    }
+    return maxModelLength;
+}
+
 Status determinePipelineType(PipelineType& pipelineType, const mediapipe::LLMCalculatorOptions& nodeOptions, const std::string& graphPath) {
     // Assuming that models_path is always set
     std::string parsedModelsPath;
@@ -278,28 +305,4 @@ Status initializeGenAiServable(std::shared_ptr<GenAiServable>& servable, const :
     }
     return StatusCode::OK;
 }
-std::optional<uint32_t> parseMaxModelLength(std::string& modelsPath) {
-    std::string configPath = FileSystem::appendSlash(modelsPath) + "config.json";
-    std::optional<uint32_t> maxModelLength;
-    if (std::filesystem::exists(configPath.c_str())) {
-        std::ifstream ifs(configPath);
-        if (!ifs.is_open()) {
-            return maxModelLength;
-        }
-        rapidjson::Document modelConfig;
-        rapidjson::IStreamWrapper isw(ifs);
-        rapidjson::ParseResult parseResult = modelConfig.ParseStream(isw);
-        if (parseResult.Code()) {
-            return maxModelLength;
-        }
-        std::vector<std::string> maxLengthFields = {"max_position_embeddings", "n_positions", "seq_len", "seq_length", "n_ctx", "sliding_window"};
-        for (auto field : maxLengthFields) {
-            if (modelConfig.HasMember(field.c_str()) && modelConfig[field.c_str()].IsUint()) {
-                maxModelLength = modelConfig[field.c_str()].GetUint();
-                break;
-            }
-        }
-    }
-    return maxModelLength;
-}
 }  // namespace ovms
diff --git a/src/llm/servable_initializer.hpp b/src/llm/servable_initializer.hpp
index ad6803d8c5..a5f46ba6cc 100644
--- a/src/llm/servable_initializer.hpp
+++ b/src/llm/servable_initializer.hpp
@@ -46,7 +46,7 @@ struct GenAiServableProperties;
 class GenAiServableInitializer {
 public:
     virtual ~GenAiServableInitializer() = default;
-    static void loadTextProcessor(std::shared_ptr<GenAiServableProperties> properties, const std::string& chatTemplateDirectory);
+    static void loadTemplateProcessor(std::shared_ptr<GenAiServableProperties> properties, const std::string& chatTemplateDirectory);
     /*
     initialize method implementation MUST fill servable with all required properties i.e. pipeline, tokenizer, configs etc. based on mediapipe node options.
     It is strictly connected with the servable, so implementation of this method in a derived class should be aware of the specific servable class structure
@@ -55,7 +55,7 @@ class GenAiServableInitializer {
     virtual Status initialize(std::shared_ptr<GenAiServable>& servable, const mediapipe::LLMCalculatorOptions& nodeOptions, std::string graphPath) = 0;
 };
 Status parseModelsPath(std::string& outPath, std::string modelsPath, std::string graphPath);
+std::optional<uint32_t> parseMaxModelLength(std::string& modelsPath);
 Status determinePipelineType(PipelineType& pipelineType, const mediapipe::LLMCalculatorOptions& nodeOptions, const std::string& graphPath);
 Status initializeGenAiServable(std::shared_ptr<GenAiServable>& servable, const ::mediapipe::CalculatorGraphConfig::Node& graphNodeConfig, std::string graphPath);
-std::optional<uint32_t> parseMaxModelLength(std::string& modelsPath);
 }  // namespace ovms
diff --git a/src/llm/text_processor.hpp b/src/llm/text_utils.hpp
similarity index 75%
rename from src/llm/text_processor.hpp
rename to src/llm/text_utils.hpp
index 4ceed95d6d..fba00730f5 100644
--- a/src/llm/text_processor.hpp
+++ b/src/llm/text_utils.hpp
@@ -1,5 +1,5 @@
 //*****************************************************************************
-// Copyright 2024 Intel Corporation
+// Copyright 2025 Intel Corporation
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -16,30 +16,9 @@
 #pragma once
 
 #include <cstdint>
-#include <memory>
 #include <sstream>
 #include <string>
-
 #include <openvino/openvino.hpp>
-#pragma warning(push)
-#pragma warning(disable : 6326 28182 6011 28020)
-// Python execution for template processing
-#include <pybind11/embed.h>  // everything needed for embedding
-#include <pybind11/stl.h>
-#pragma warning(pop)
-
-#include "src/python/utils.hpp"
-
-namespace ovms {
-
-class TextProcessor {
-public:
-    std::string bosToken = "";
-    std::string eosToken = "";
-    std::unique_ptr<PyObjectWrapper<py::object>> chatTemplate = nullptr;
-
-    static bool applyChatTemplate(TextProcessor& textProcessor, std::string modelsPath, const std::string& requestBody, std::string& output);
-};
 
 template <typename T>
 static std::string packPromptTokens(T* input, size_t size) {
@@ -81,4 +60,3 @@ static std::string getPromptTokensString(const ov::Tensor& tensor) {
 }
 #pragma GCC diagnostic pop
 #pragma warning(pop)
-}  // namespace ovms
diff --git a/src/llm/visual_language_model/continuous_batching/servable.cpp b/src/llm/visual_language_model/continuous_batching/servable.cpp
index 515e2c2395..e876690a7a 100644
--- a/src/llm/visual_language_model/continuous_batching/servable.cpp
+++ b/src/llm/visual_language_model/continuous_batching/servable.cpp
@@ -23,6 +23,7 @@
 #include <vector>
 
 #include "../../../logging.hpp"
+#include "../../text_utils.hpp"
 
 namespace ovms {
 
diff --git a/src/llm/visual_language_model/legacy/servable.cpp b/src/llm/visual_language_model/legacy/servable.cpp
index 75b7c42d25..af29e4a1e7 100644
--- a/src/llm/visual_language_model/legacy/servable.cpp
+++ b/src/llm/visual_language_model/legacy/servable.cpp
@@ -34,7 +34,10 @@
 #include "../../../http_payload.hpp"
 #include "../../../mediapipe_internal/mediapipe_utils.hpp"
 #include "../../apis/openai_completions.hpp"
-#include "../../text_processor.hpp"
+#include "../../text_utils.hpp"
+#if (PYTHON_DISABLE == 0)
+#include "../../py_jinja_template_processor.hpp"
+#endif
 #include "servable.hpp"
 
 namespace ovms {
diff --git a/src/llm/visual_language_model/legacy/servable_initializer.cpp b/src/llm/visual_language_model/legacy/servable_initializer.cpp
index ba2e1df5c3..c10bf1b3af 100644
--- a/src/llm/visual_language_model/legacy/servable_initializer.cpp
+++ b/src/llm/visual_language_model/legacy/servable_initializer.cpp
@@ -81,8 +81,9 @@ Status VisualLanguageModelLegacyServableInitializer::initialize(std::shared_ptr<
         SPDLOG_ERROR("Error during llm node initialization for models_path: {}", parsedModelsPath);
         return StatusCode::LLM_NODE_RESOURCE_STATE_INITIALIZATION_FAILED;
     }
-
-    loadTextProcessor(properties, parsedModelsPath);
+#if (PYTHON_DISABLE == 0)
+    loadTemplateProcessor(properties, parsedModelsPath);
+#endif
     properties->legacyExecutor = std::make_shared<VisualLanguageModelLegacyExecutorWrapper>(properties->pipeline);
     if (nodeOptions.has_max_tokens_limit()) {
         properties->maxTokensLimit = nodeOptions.max_tokens_limit();
diff --git a/src/mediapipe_internal/mediapipegraphdefinition.cpp b/src/mediapipe_internal/mediapipegraphdefinition.cpp
index d3bd6b8bb4..dd08768f3a 100644
--- a/src/mediapipe_internal/mediapipegraphdefinition.cpp
+++ b/src/mediapipe_internal/mediapipegraphdefinition.cpp
@@ -33,9 +33,9 @@
 #include "../model_metric_reporter.hpp"
 #include "../modelmanager.hpp"
 #include "../ov_utils.hpp"
-#if (PYTHON_DISABLE == 0)
 #include "../llm/servable.hpp"
 #include "../llm/servable_initializer.hpp"
+#if (PYTHON_DISABLE == 0)
 #include "../python/pythonnoderesources.hpp"
 #endif
 #include "../status.hpp"
@@ -396,7 +396,6 @@ Status MediapipeGraphDefinition::waitForLoaded(std::unique_ptr<MediapipeGraphDef
     return StatusCode::OK;
 }
 
-#if (PYTHON_DISABLE == 0)
 template <typename T>
 class ResourcesCleaningGuard {
 public:
@@ -413,7 +412,6 @@ class ResourcesCleaningGuard {
         shouldCleanup = false;
     }
 };
-#endif
 
 Status MediapipeGraphDefinition::initializeNodes() {
     SPDLOG_INFO("MediapipeGraphDefinition initializing graph nodes");
@@ -445,6 +443,7 @@ Status MediapipeGraphDefinition::initializeNodes() {
             this->pythonNodeResourcesMap.insert(std::pair<std::string, std::shared_ptr<PythonNodeResources>>(nodeName, std::move(nodeResources)));
             pythonResourcesCleaningGuard.disableCleaning();
         }
+#endif
         // Passed to both calculators that require LLM Engine (gRPC KServe & HTTP OpenAI)
         if (endsWith(config.node(i).calculator(), LLM_NODE_CALCULATOR_NAME)) {
             ResourcesCleaningGuard<GenAiServableMap> genAiServablesCleaningGuard(this->genAiServableMap);
@@ -470,7 +469,6 @@ Status MediapipeGraphDefinition::initializeNodes() {
             this->genAiServableMap.insert(std::pair<std::string, std::shared_ptr<GenAiServable>>(nodeName, std::move(servable)));
             genAiServablesCleaningGuard.disableCleaning();
         }
-#endif
     }
     return StatusCode::OK;
 }
diff --git a/src/mediapipe_internal/mediapipegraphexecutor.hpp b/src/mediapipe_internal/mediapipegraphexecutor.hpp
index c785adb9a8..bd2fdf5134 100644
--- a/src/mediapipe_internal/mediapipegraphexecutor.hpp
+++ b/src/mediapipe_internal/mediapipegraphexecutor.hpp
@@ -134,8 +134,8 @@ class MediapipeGraphExecutor {
         OVMS_RETURN_ON_FAIL(deserializeInputSidePacketsFromFirstRequestImpl(inputSidePackets, *request));
 #if (PYTHON_DISABLE == 0)
         inputSidePackets[PYTHON_SESSION_SIDE_PACKET_TAG] = mediapipe::MakePacket<PythonNodeResourcesMap>(this->pythonNodeResourcesMap).At(STARTING_TIMESTAMP);
-        inputSidePackets[LLM_SESSION_SIDE_PACKET_TAG] = mediapipe::MakePacket<GenAiServableMap>(this->llmNodeResourcesMap).At(STARTING_TIMESTAMP);
 #endif
+        inputSidePackets[LLM_SESSION_SIDE_PACKET_TAG] = mediapipe::MakePacket<GenAiServableMap>(this->llmNodeResourcesMap).At(STARTING_TIMESTAMP);
         MP_RETURN_ON_FAIL(graph.StartRun(inputSidePackets), std::string("start MediaPipe graph: ") + this->name, StatusCode::MEDIAPIPE_GRAPH_START_ERROR);
 
         ::mediapipe::Packet packet;
@@ -278,8 +278,8 @@ class MediapipeGraphExecutor {
 #if (PYTHON_DISABLE == 0)
                 inputSidePackets[PYTHON_SESSION_SIDE_PACKET_TAG] = mediapipe::MakePacket<PythonNodeResourcesMap>(this->pythonNodeResourcesMap)
                                                                        .At(STARTING_TIMESTAMP);
-                inputSidePackets[LLM_SESSION_SIDE_PACKET_TAG] = mediapipe::MakePacket<GenAiServableMap>(this->llmNodeResourcesMap).At(STARTING_TIMESTAMP);
 #endif
+                inputSidePackets[LLM_SESSION_SIDE_PACKET_TAG] = mediapipe::MakePacket<GenAiServableMap>(this->llmNodeResourcesMap).At(STARTING_TIMESTAMP);
             }
 
             {
diff --git a/src/test/llm/llmnode_test.cpp b/src/test/llm/llmnode_test.cpp
index da2b967b82..0d5a746d51 100644
--- a/src/test/llm/llmnode_test.cpp
+++ b/src/test/llm/llmnode_test.cpp
@@ -26,10 +26,12 @@
 #include <gtest/gtest.h>
 #include <openvino/genai/continuous_batching_pipeline.hpp>
 #include <openvino/openvino.hpp>
+#if (PYTHON_DISABLE == 0)
 #pragma warning(push)
 #pragma warning(disable : 6326 28182 6011 28020)
 #include <pybind11/embed.h>
 #pragma warning(pop)
+#endif
 
 #include "../../http_rest_api_handler.hpp"
 #include "../../http_status_code.hpp"
@@ -39,7 +41,7 @@
 #include "../../llm/language_model/continuous_batching/servable.hpp"
 #include "../../llm/servable.hpp"
 #include "../../llm/servable_initializer.hpp"
-#include "../../llm/text_processor.hpp"
+#include "../../llm/text_utils.hpp"
 #include "../../ov_utils.hpp"
 #include "../../server.hpp"
 #include "rapidjson/document.h"
@@ -3235,9 +3237,11 @@ INSTANTIATE_TEST_SUITE_P(
 
 // Common tests for all pipeline types (testing logic executed prior pipeline type selection)
 class LLMConfigHttpTest : public ::testing::Test {
+#if (PYTHON_DISABLE == 0)
 public:
     void SetUp() { py::initialize_interpreter(); }
     void TearDown() { py::finalize_interpreter(); }
+#endif
 };
 
 TEST_F(LLMConfigHttpTest, LLMNodeNameMissing) {
@@ -3494,9 +3498,11 @@ TEST_F(LLMConfigHttpTest, LLMNodeWorkspacePathToFileNotDir) {
 }
 
 class LLMConfigHttpTestParameterized : public ::testing::Test, public ::testing::WithParamInterface<std::tuple<std::string, ovms::StatusCode>> {
+#if (PYTHON_DISABLE == 0)
 public:
     void SetUp() { py::initialize_interpreter(); }
     void TearDown() { py::finalize_interpreter(); }
+#endif
 };
 
 TEST_P(LLMConfigHttpTestParameterized, LLMNodeResourceInitFailed) {
@@ -3560,9 +3566,11 @@ INSTANTIATE_TEST_SUITE_P(
 // Those tests are working on Continuous Batching path, since most of the node options are scheduler parameters that are not used in non-CB servables
 // We could consider adding tests for non-CB path in the future in the separate test suite
 class LLMOptionsHttpTestPython : public ::testing::Test {
+#if (PYTHON_DISABLE == 0)
 public:
     static void SetUpTestSuite() { py::initialize_interpreter(); }
     static void TearDownTestSuite() { py::finalize_interpreter(); }
+#endif
 };
 
 class LLMOptionsHttpTest : public LLMOptionsHttpTestPython {
diff --git a/src/test/llm/llmtemplate_test.cpp b/src/test/llm/llmtemplate_test.cpp
index 86d1bb24bd..752dec7fbc 100644
--- a/src/test/llm/llmtemplate_test.cpp
+++ b/src/test/llm/llmtemplate_test.cpp
@@ -33,7 +33,7 @@
 #include "../../httpservermodule.hpp"
 #include "../../llm/language_model/continuous_batching/servable.hpp"
 #include "../../llm/language_model/continuous_batching/servable_initializer.hpp"
-#include "../../llm/text_processor.hpp"
+#include "../../llm/py_jinja_template_processor.hpp"
 #include "../../mediapipe_internal/mediapipegraphdefinition.hpp"
 #include "../../server.hpp"
 
@@ -77,11 +77,11 @@ TEST_F(LLMChatTemplateTest, ChatTemplateEmptyBody) {
 
     servable->getProperties()->modelsPath = directoryPath;
     // default_chat_template = "{% if messages|length != 1 %} {{ raise_exception('This servable accepts only single message requests') }}{% endif %}{{ messages[0]['content'] }}"
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = "";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
     std::string errorOutput = "Expecting value: line 1 column 1 (char 0)";
     ASSERT_EQ(finalPrompt, errorOutput);
 }
@@ -91,7 +91,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateEmptyMessage) {
 
     servable->getProperties()->modelsPath = directoryPath;
     // default_chat_template = "{% if messages|length != 1 %} {{ raise_exception('This servable accepts only single message requests') }}{% endif %}{{ messages[0]['content'] }}"
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -102,7 +102,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateEmptyMessage) {
         }
     )";
     std::string errorOutput = "This servable accepts only single message requests";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
     ASSERT_EQ(finalPrompt, errorOutput);
 }
 
@@ -111,7 +111,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateMessageWithEmptyObject) {
 
     servable->getProperties()->modelsPath = directoryPath;
     // default_chat_template = "{% if messages|length != 1 %} {{ raise_exception('This servable accepts only single message requests') }}{% endif %}{{ messages[0]['content'] }}"
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -121,7 +121,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateMessageWithEmptyObject) {
             "messages": [{}]
         }
     )";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
     ASSERT_EQ(finalPrompt, "");
 }
 
@@ -130,7 +130,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateDefault) {
 
     servable->getProperties()->modelsPath = directoryPath;
     // default_chat_template = "{% if messages|length != 1 %} {{ raise_exception('This servable accepts only single message requests') }}{% endif %}{{ messages[0]['content'] }}"
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -139,7 +139,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateDefault) {
         }
     )";
     std::string expectedOutput = "How can I help you?";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
@@ -148,7 +148,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateMultiMessage) {
 
     servable->getProperties()->modelsPath = directoryPath;
     // default_chat_template = "{% if messages|length != 1 %} {{ raise_exception('This servable accepts only single message requests') }}{% endif %}{{ messages[0]['content'] }}"
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -157,7 +157,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateMultiMessage) {
         }
     )";
     std::string errorOutput = "This servable accepts only single message requests";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
     ASSERT_EQ(finalPrompt, errorOutput);
 }
 
@@ -166,7 +166,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateComplexMessage) {
 
     servable->getProperties()->modelsPath = directoryPath;
     // default_chat_template = "{% if messages|length != 1 %} {{ raise_exception('This servable accepts only single message requests') }}{% endif %}{{ messages[0]['content'] }}"
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -177,7 +177,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateComplexMessage) {
         }
     )";
     std::string expectedOutput = "hello";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
@@ -187,7 +187,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateJinjaUppercase) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -198,7 +198,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateJinjaUppercase) {
         }
     )";
     std::string expectedOutput = " Hi, HELLO ";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
@@ -208,7 +208,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateJinjaException) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -219,7 +219,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateJinjaException) {
         }
     )";
     std::string errorOutput = "list object has no element 3";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
     ASSERT_EQ(finalPrompt, errorOutput);
 }
 
@@ -232,7 +232,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerDefault) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -243,7 +243,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerDefault) {
         }
     )";
     std::string expectedOutput = "hello";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
@@ -256,7 +256,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerBosNull) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -268,7 +268,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerBosNull) {
     )";
     std::string expectedOutput = "hello";
     // Expect no issues with chat template since non string bos token is ignored
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
@@ -281,7 +281,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerBosDict) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -293,7 +293,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerBosDict) {
     )";
     std::string expectedError = "Error: Chat template not loaded correctly, so it cannot be applied";
     // Expect no issues with chat template since non string bos token is ignored
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
     ASSERT_EQ(finalPrompt, expectedError);
 }
 
@@ -306,7 +306,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerEosNull) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -318,7 +318,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerEosNull) {
     )";
     std::string expectedOutput = "hello";
     // Expect no issues with chat template since non string eos token is ignored
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
@@ -331,7 +331,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerException) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -342,7 +342,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerException) {
         }
     )";
     std::string expectedOutput = "Error: Chat template not loaded correctly, so it cannot be applied";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
@@ -356,7 +356,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerUpperCase) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -367,7 +367,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerUpperCase) {
         }
     )";
     std::string expectedOutput = "Hi, HELLO";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
@@ -381,7 +381,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerTemplateException) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -392,7 +392,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerTemplateException) {
         }
     )";
     std::string expectedOutput = "list object has no element 3";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
@@ -406,7 +406,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerTemplateBadVariable) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -417,7 +417,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTokenizerTemplateBadVariable) {
         }
     )";
     std::string expectedError = "Error: Chat template not loaded correctly, so it cannot be applied";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), false);
     ASSERT_EQ(finalPrompt, expectedError);
 }
 
@@ -434,7 +434,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTwoConfigs) {
     std::shared_ptr<GenAiServable> servable = std::make_shared<ContinuousBatchingServable>();
 
     servable->getProperties()->modelsPath = directoryPath;
-    GenAiServableInitializer::loadTextProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
+    GenAiServableInitializer::loadTemplateProcessor(servable->getProperties(), servable->getProperties()->modelsPath);
 
     std::string finalPrompt = "";
     std::string payloadBody = R"(
@@ -445,7 +445,7 @@ TEST_F(LLMChatTemplateTest, ChatTemplateTwoConfigs) {
         }
     )";
     std::string expectedOutput = " Hi, HELLO ";
-    ASSERT_EQ(TextProcessor::applyChatTemplate(servable->getProperties()->textProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
+    ASSERT_EQ(PyJinjaTemplateProcessor::applyChatTemplate(servable->getProperties()->templateProcessor, servable->getProperties()->modelsPath, payloadBody, finalPrompt), true);
     ASSERT_EQ(finalPrompt, expectedOutput);
 }
 
diff --git a/src/test/llm/text_streamer_test.cpp b/src/test/llm/text_streamer_test.cpp
index 9905db756e..386821f4b3 100644
--- a/src/test/llm/text_streamer_test.cpp
+++ b/src/test/llm/text_streamer_test.cpp
@@ -40,7 +40,9 @@ class TextStreamerTest : public ::testing::Test {
 )";
 
     static void SetUpTestSuite() {
+#if (PYTHON_DISABLE == 0)
         py::initialize_interpreter();
+#endif
         std::string adjustedPbtxt = testPbtxt;
         adjustConfigForTargetPlatform(adjustedPbtxt);
         ASSERT_TRUE(::google::protobuf::TextFormat::ParseFromString(adjustedPbtxt, &config));
@@ -54,7 +56,9 @@ class TextStreamerTest : public ::testing::Test {
     }
     static void TearDownTestSuite() {
         servable.reset();
+#if (PYTHON_DISABLE == 0)
         py::finalize_interpreter();
+#endif
     }
     void assertTokensValues(ov::Tensor generatedTokens, std::vector<int64_t> expectedTokens) {
         ASSERT_EQ(generatedTokens.get_size(), expectedTokens.size());
diff --git a/src/test/llm/visual_language_model/initialization_test.cpp b/src/test/llm/visual_language_model/initialization_test.cpp
index a3c5ca2224..ee31677886 100644
--- a/src/test/llm/visual_language_model/initialization_test.cpp
+++ b/src/test/llm/visual_language_model/initialization_test.cpp
@@ -34,9 +34,11 @@ Status callDeterminePipelineType(PipelineType& pipelineType, const std::string&
 
 // Initialization tests
 class VLMServableInitializationTest : public ::testing::Test {
+#if (PYTHON_DISABLE == 0)
 public:
     void SetUp() { py::initialize_interpreter(); }
     void TearDown() { py::finalize_interpreter(); }
+#endif
 };
 
 TEST_F(VLMServableInitializationTest, determinePipelineTypeDefault) {
diff --git a/src/test/mediapipeflow_test.cpp b/src/test/mediapipeflow_test.cpp
index 8f67aecb60..bf814ec0bb 100644
--- a/src/test/mediapipeflow_test.cpp
+++ b/src/test/mediapipeflow_test.cpp
@@ -3667,8 +3667,8 @@ TEST(WhitelistRegistered, MediapipeCalculatorsList) {
         "CalculatorRunnerSourceCalculator",
         "PyTensorOvTensorConverterCalculator",   // integral OVMS calculator
         "PythonExecutorCalculator",  // integral OVMS calculator
-        "HttpLLMCalculator",  // integral OVMS calculator
 #endif
+        "HttpLLMCalculator",  // integral OVMS calculator
         "OpenAIChatCompletionsMockCalculator",  // OVMS test calculator
         "AddHeaderCalculator",
         "AddNumbersMultiInputsOutputsTestCalculator",