diff --git a/src/plugins/auto_batch/src/async_infer_request.cpp b/src/plugins/auto_batch/src/async_infer_request.cpp
index a92e8c7f06ff43..d740d9d5861ed6 100644
--- a/src/plugins/auto_batch/src/async_infer_request.cpp
+++ b/src/plugins/auto_batch/src/async_infer_request.cpp
@@ -69,6 +69,7 @@ AsyncInferRequest::AsyncInferRequest(const std::shared_ptr<SyncInferRequest>& re
                 // it is ok to call size() here as the queue only grows (and the bulk removal happens under the mutex)
                 const int sz = static_cast<int>(workerInferRequest->_tasks.size());
                 if (sz == workerInferRequest->_batch_size) {
+                    workerInferRequest->_is_wakeup = true;
                     workerInferRequest->_cond.notify_one();
                 }
             };
diff --git a/src/plugins/auto_batch/src/compiled_model.cpp b/src/plugins/auto_batch/src/compiled_model.cpp
index 68de2481a3e019..b02759f53dc1d1 100644
--- a/src/plugins/auto_batch/src/compiled_model.cpp
+++ b/src/plugins/auto_batch/src/compiled_model.cpp
@@ -64,6 +64,7 @@ CompiledModel::GetWorkerInferRequest() const {
             workerRequestPtr->_infer_request_batched._so = m_compiled_model_with_batch._so;
         workerRequestPtr->_batch_size = m_device_info.device_batch_size;
         workerRequestPtr->_completion_tasks.resize(workerRequestPtr->_batch_size);
+        workerRequestPtr->_is_wakeup = false;
         workerRequestPtr->_infer_request_batched->set_callback(
             [workerRequestPtr](std::exception_ptr exceptionPtr) mutable {
                 if (exceptionPtr)
@@ -74,6 +75,7 @@ CompiledModel::GetWorkerInferRequest() const {
                     workerRequestPtr->_completion_tasks[c]();
                 }
                 // reset the timeout
+                workerRequestPtr->_is_wakeup = true;
                 workerRequestPtr->_cond.notify_one();
             });
 
@@ -83,6 +85,9 @@ CompiledModel::GetWorkerInferRequest() const {
                 {
                     std::unique_lock<std::mutex> lock(workerRequestPtr->_mutex);
                     status = workerRequestPtr->_cond.wait_for(lock, std::chrono::milliseconds(m_time_out));
+                    if ((status != std::cv_status::timeout) && (workerRequestPtr->_is_wakeup == false))
+                        continue;
+                    workerRequestPtr->_is_wakeup = false;
                 }
                 if (m_terminate) {
                     break;
diff --git a/src/plugins/auto_batch/src/compiled_model.hpp b/src/plugins/auto_batch/src/compiled_model.hpp
index 200445f0140495..c4cf38d0e701ea 100644
--- a/src/plugins/auto_batch/src/compiled_model.hpp
+++ b/src/plugins/auto_batch/src/compiled_model.hpp
@@ -30,6 +30,7 @@ class CompiledModel : public ov::ICompiledModel {
         std::condition_variable _cond;
         std::mutex _mutex;
         std::exception_ptr _exception_ptr;
+        bool _is_wakeup;
     };
 
     CompiledModel(const std::shared_ptr<ov::Model>& model,