diff --git a/llama_stack/providers/inline/agents/meta_reference/responses/openai_responses.py b/llama_stack/providers/inline/agents/meta_reference/responses/openai_responses.py
index aa04f1afa..2360dafd9 100644
--- a/llama_stack/providers/inline/agents/meta_reference/responses/openai_responses.py
+++ b/llama_stack/providers/inline/agents/meta_reference/responses/openai_responses.py
@@ -123,19 +123,6 @@ class OpenAIResponsesImpl:
                 # Use stored messages directly and convert only new input
                 message_adapter = TypeAdapter(list[OpenAIMessageParam])
                 messages = message_adapter.validate_python(previous_response.messages)
-                # When managing conversation state with the previous_response_id parameter,
-                # the instructions used on previous turns will not be carried over in the context
-                previous_instructions = previous_response.instructions
-                if previous_instructions:
-                    if (
-                        isinstance(previous_instructions, str)
-                        and previous_instructions == messages[0].content
-                        and messages[0].role == "system"
-                    ):
-                        # Omit instructions from previous response
-                        del messages[0]
-                    else:
-                        raise ValueError("Instructions from the previous response could not be validated")
                 new_messages = await convert_response_input_to_chat_messages(input, previous_messages=messages)
                 messages.extend(new_messages)
             else:
diff --git a/tests/unit/providers/agents/meta_reference/test_openai_responses.py b/tests/unit/providers/agents/meta_reference/test_openai_responses.py
index 5e95439f5..54c1820fb 100644
--- a/tests/unit/providers/agents/meta_reference/test_openai_responses.py
+++ b/tests/unit/providers/agents/meta_reference/test_openai_responses.py
@@ -38,7 +38,6 @@ from llama_stack.apis.inference import (
     OpenAIJSONSchema,
     OpenAIResponseFormatJSONObject,
     OpenAIResponseFormatJSONSchema,
-    OpenAISystemMessageParam,
     OpenAIUserMessageParam,
 )
 from llama_stack.apis.tools.tools import ListToolDefsResponse, ToolDef, ToolGroups, ToolInvocationResult, ToolRuntime
@@ -840,7 +839,6 @@ async def test_create_openai_response_with_previous_response_instructions(
         text=OpenAIResponseText(format=OpenAIResponseTextFormat(type="text")),
         input=[input_item_message],
         messages=[
-            OpenAISystemMessageParam(content="You are a helpful assistant."),
             OpenAIUserMessageParam(content="Name some towns in Ireland"),
             OpenAIAssistantMessageParam(content="Galway, Longford, Sligo"),
         ],