diff --git a/backend/agentpress/response_processor.py b/backend/agentpress/response_processor.py
index 40eb6675..470aaf44 100644
--- a/backend/agentpress/response_processor.py
+++ b/backend/agentpress/response_processor.py
@@ -798,16 +798,20 @@ class ResponseProcessor:
             logger.error(f"Error processing stream: {str(e)}", exc_info=True)
             self.trace.event(name="error_processing_stream", level="ERROR", status_message=(f"Error processing stream: {str(e)}"))
             # Save and yield error status message
-            err_content = {"role": "system", "status_type": "error", "message": str(e)}
-            err_msg_obj = await self.add_message(
-                thread_id=thread_id, type="status", content=err_content, 
-                is_llm_message=False, metadata={"thread_run_id": thread_run_id if 'thread_run_id' in locals() else None}
-            )
-            if err_msg_obj: yield format_for_yield(err_msg_obj) # Yield the saved error message
             
-            # Re-raise the same exception (not a new one) to ensure proper error propagation
-            logger.critical(f"Re-raising error to stop further processing: {str(e)}")
-            self.trace.event(name="re_raising_error_to_stop_further_processing", level="ERROR", status_message=(f"Re-raising error to stop further processing: {str(e)}"))
+            err_content = {"role": "system", "status_type": "error", "message": str(e)}
+            if (not "AnthropicException - Overloaded" in str(e)):
+                err_msg_obj = await self.add_message(
+                    thread_id=thread_id, type="status", content=err_content, 
+                    is_llm_message=False, metadata={"thread_run_id": thread_run_id if 'thread_run_id' in locals() else None}
+                )
+                if err_msg_obj: yield format_for_yield(err_msg_obj) # Yield the saved error message
+                # Re-raise the same exception (not a new one) to ensure proper error propagation
+                logger.critical(f"Re-raising error to stop further processing: {str(e)}")
+                self.trace.event(name="re_raising_error_to_stop_further_processing", level="ERROR", status_message=(f"Re-raising error to stop further processing: {str(e)}"))
+            else:
+                logger.error(f"AnthropicException - Overloaded detected - Falling back to OpenRouter: {str(e)}", exc_info=True)
+                self.trace.event(name="anthropic_exception_overloaded_detected", level="ERROR", status_message=(f"AnthropicException - Overloaded detected - Falling back to OpenRouter: {str(e)}"))
             raise # Use bare 'raise' to preserve the original exception with its traceback
 
         finally:
diff --git a/backend/agentpress/thread_manager.py b/backend/agentpress/thread_manager.py
index ec46710e..a4d1d974 100644
--- a/backend/agentpress/thread_manager.py
+++ b/backend/agentpress/thread_manager.py
@@ -477,13 +477,20 @@ Here are the XML tools available with examples:
                         if not auto_continue:
                             break
                     except Exception as e:
-                        # If there's an exception, log it, yield an error status, and stop execution
-                        logger.error(f"Error in auto_continue_wrapper generator: {str(e)}", exc_info=True)
-                        yield {
-                            "type": "status",
-                            "status": "error",
-                            "message": f"Error in thread processing: {str(e)}"
-                        }
+                        if ("AnthropicException - Overloaded" in str(e)):
+                            logger.error(f"AnthropicException - Overloaded detected - Falling back to OpenRouter: {str(e)}", exc_info=True)
+                            nonlocal llm_model
+                            llm_model = f"openrouter/{llm_model}"
+                            auto_continue = True
+                            continue # Continue the loop
+                        else:
+                            # If there's any other exception, log it, yield an error status, and stop execution
+                            logger.error(f"Error in auto_continue_wrapper generator: {str(e)}", exc_info=True)
+                            yield {
+                                "type": "status",
+                                "status": "error",
+                                "message": f"Error in thread processing: {str(e)}"
+                            }
                         return  # Exit the generator on any error
                 except Exception as outer_e:
                     # Catch exceptions from _run_once itself
diff --git a/backend/services/llm.py b/backend/services/llm.py
index 1b5c61c5..f63f4539 100644
--- a/backend/services/llm.py
+++ b/backend/services/llm.py
@@ -172,7 +172,7 @@ def prepare_params(
 
         # Apply cache control to the first 4 text blocks across all messages
         cache_control_count = 0
-        max_cache_control_blocks = 4
+        max_cache_control_blocks = 3
 
         for message in messages:
             if cache_control_count >= max_cache_control_blocks:
diff --git a/backend/utils/logger.py b/backend/utils/logger.py
index 4c9d805f..2d7ecf14 100644
--- a/backend/utils/logger.py
+++ b/backend/utils/logger.py
@@ -6,8 +6,8 @@ LOGGING_LEVEL = logging.getLevelNamesMapping().get(
 )
 
 renderer = [structlog.processors.JSONRenderer()]
-if ENV_MODE.lower() == "local".lower() or ENV_MODE.lower() == "staging".lower():
-    renderer = [structlog.dev.ConsoleRenderer()]
+# if ENV_MODE.lower() == "local".lower() or ENV_MODE.lower() == "staging".lower():
+#     renderer = [structlog.dev.ConsoleRenderer()]
 
 structlog.configure(
     processors=[