diff --git a/python/packages/azure-ai/agent_framework_azure_ai/_client.py b/python/packages/azure-ai/agent_framework_azure_ai/_client.py
index b06b7b5df0..2e8edc7e47 100644
--- a/python/packages/azure-ai/agent_framework_azure_ai/_client.py
+++ b/python/packages/azure-ai/agent_framework_azure_ai/_client.py
@@ -300,13 +300,26 @@ def _create_text_format_config(
         raise ServiceInvalidRequestError("response_format must be a Pydantic model or mapping.")
 
     async def _get_agent_reference_or_create(
-        self, run_options: dict[str, Any], messages_instructions: str | None
+        self,
+        run_options: dict[str, Any],
+        messages_instructions: str | None,
+        chat_options: ChatOptions | None = None,
     ) -> dict[str, str]:
         """Determine which agent to use and create if needed.
 
+        Args:
+            run_options: The prepared options for the API call.
+            messages_instructions: Instructions extracted from messages.
+            chat_options: The chat options containing response_format and other settings.
+
         Returns:
             dict[str, str]: The agent reference to use.
         """
+        # chat_options is needed separately because the base class excludes response_format
+        # from run_options (transforming it to text/text_format for OpenAI). Azure's agent
+        # creation API requires the original response_format to build its own config format.
+        if chat_options is None:
+            chat_options = ChatOptions()
         # Agent name must be explicitly provided by the user.
         if self.agent_name is None:
             raise ServiceInitializationError(
@@ -341,8 +354,14 @@ async def _get_agent_reference_or_create(
             if "top_p" in run_options:
                 args["top_p"] = run_options["top_p"]
 
-            if "response_format" in run_options:
-                response_format = run_options["response_format"]
+            # response_format is accessed from chat_options or additional_properties
+            # since the base class excludes it from run_options
+            response_format: Any = (
+                chat_options.response_format
+                if chat_options.response_format is not None
+                else chat_options.additional_properties.get("response_format")
+            )
+            if response_format:
                 args["text"] = PromptAgentDefinitionText(format=self._create_text_format_config(response_format))
 
             # Combine instructions from messages and options
@@ -390,12 +409,12 @@ async def _prepare_options(
 
         if not self._is_application_endpoint:
             # Application-scoped response APIs do not support "agent" property.
-            agent_reference = await self._get_agent_reference_or_create(run_options, instructions)
+            agent_reference = await self._get_agent_reference_or_create(run_options, instructions, chat_options)
             run_options["extra_body"] = {"agent": agent_reference}
 
         # Remove properties that are not supported on request level
         # but were configured on agent level
-        exclude = ["model", "tools", "response_format", "temperature", "top_p"]
+        exclude = ["model", "tools", "response_format", "temperature", "top_p", "text", "text_format"]
 
         for property in exclude:
             run_options.pop(property, None)
diff --git a/python/packages/azure-ai/tests/test_azure_ai_client.py b/python/packages/azure-ai/tests/test_azure_ai_client.py
index 2ca49c2033..3b1b500ede 100644
--- a/python/packages/azure-ai/tests/test_azure_ai_client.py
+++ b/python/packages/azure-ai/tests/test_azure_ai_client.py
@@ -723,9 +723,10 @@ async def test_azure_ai_client_agent_creation_with_response_format(
     mock_agent.version = "1.0"
     mock_project_client.agents.create_version = AsyncMock(return_value=mock_agent)
 
-    run_options = {"model": "test-model", "response_format": ResponseFormatModel}
+    run_options = {"model": "test-model"}
+    chat_options = ChatOptions(response_format=ResponseFormatModel)
 
-    await client._get_agent_reference_or_create(run_options, None)  # type: ignore
+    await client._get_agent_reference_or_create(run_options, None, chat_options)  # type: ignore
 
     # Verify agent was created with response format configuration
     call_args = mock_project_client.agents.create_version.call_args
@@ -776,19 +777,18 @@ async def test_azure_ai_client_agent_creation_with_mapping_response_format(
         "additionalProperties": False,
     }
 
-    run_options = {
-        "model": "test-model",
-        "response_format": {
-            "type": "json_schema",
-            "json_schema": {
-                "name": runtime_schema["title"],
-                "strict": True,
-                "schema": runtime_schema,
-            },
+    run_options = {"model": "test-model"}
+    response_format_mapping = {
+        "type": "json_schema",
+        "json_schema": {
+            "name": runtime_schema["title"],
+            "strict": True,
+            "schema": runtime_schema,
         },
     }
+    chat_options = ChatOptions(response_format=response_format_mapping)  # type: ignore
 
-    await client._get_agent_reference_or_create(run_options, None)  # type: ignore
+    await client._get_agent_reference_or_create(run_options, None, chat_options)  # type: ignore
 
     call_args = mock_project_client.agents.create_version.call_args
     created_definition = call_args[1]["definition"]
@@ -805,7 +805,7 @@ async def test_azure_ai_client_agent_creation_with_mapping_response_format(
 async def test_azure_ai_client_prepare_options_excludes_response_format(
     mock_project_client: MagicMock,
 ) -> None:
-    """Test that prepare_options excludes response_format from final run options."""
+    """Test that prepare_options excludes response_format, text, and text_format from final run options."""
     client = create_test_azure_ai_client(mock_project_client, agent_name="test-agent", agent_version="1.0")
 
     messages = [ChatMessage(role=Role.USER, contents=[TextContent(text="Hello")])]
@@ -815,7 +815,12 @@ async def test_azure_ai_client_prepare_options_excludes_response_format(
         patch.object(
             client.__class__.__bases__[0],
             "_prepare_options",
-            return_value={"model": "test-model", "response_format": ResponseFormatModel},
+            return_value={
+                "model": "test-model",
+                "response_format": ResponseFormatModel,
+                "text": {"format": {"type": "json_schema", "name": "test"}},
+                "text_format": ResponseFormatModel,
+            },
         ),
         patch.object(
             client,
@@ -825,8 +830,11 @@ async def test_azure_ai_client_prepare_options_excludes_response_format(
     ):
         run_options = await client._prepare_options(messages, chat_options)
 
-        # response_format should be excluded from final run options
+        # response_format, text, and text_format should be excluded from final run options
+        # because they are configured at agent level, not request level
         assert "response_format" not in run_options
+        assert "text" not in run_options
+        assert "text_format" not in run_options
         # But extra_body should contain agent reference
         assert "extra_body" in run_options
         assert run_options["extra_body"]["agent"]["name"] == "test-agent"
@@ -1009,3 +1017,91 @@ async def test_azure_ai_chat_client_agent_with_tools() -> None:
         assert response.text is not None
         assert len(response.text) > 0
         assert any(word in response.text.lower() for word in ["sunny", "25"])
+
+
+class ReleaseBrief(BaseModel):
+    """Structured output model for release brief."""
+
+    title: str = Field(description="A short title for the release.")
+    summary: str = Field(description="A brief summary of what was released.")
+    highlights: list[str] = Field(description="Key highlights from the release.")
+    model_config = ConfigDict(extra="forbid")
+
+
+@pytest.mark.flaky
+@skip_if_azure_ai_integration_tests_disabled
+async def test_azure_ai_chat_client_agent_with_response_format() -> None:
+    """Test ChatAgent with response_format (structured output) using AzureAIClient."""
+    async with (
+        temporary_chat_client(agent_name="ResponseFormatAgent") as chat_client,
+        ChatAgent(chat_client=chat_client) as agent,
+    ):
+        response = await agent.run(
+            "Summarize the following release notes into a ReleaseBrief:\n\n"
+            "Version 2.0 Release Notes:\n"
+            "- Added new streaming API for real-time responses\n"
+            "- Improved error handling with detailed messages\n"
+            "- Performance boost of 50% in batch processing\n"
+            "- Fixed memory leak in connection pooling",
+            response_format=ReleaseBrief,
+        )
+
+        # Validate response
+        assert isinstance(response, AgentRunResponse)
+        assert response.value is not None
+        assert isinstance(response.value, ReleaseBrief)
+
+        # Validate structured output fields
+        brief = response.value
+        assert len(brief.title) > 0
+        assert len(brief.summary) > 0
+        assert len(brief.highlights) > 0
+
+
+@pytest.mark.flaky
+@skip_if_azure_ai_integration_tests_disabled
+async def test_azure_ai_chat_client_agent_with_runtime_json_schema() -> None:
+    """Test ChatAgent with runtime JSON schema (structured output) using AzureAIClient."""
+    runtime_schema = {
+        "title": "WeatherDigest",
+        "type": "object",
+        "properties": {
+            "location": {"type": "string"},
+            "conditions": {"type": "string"},
+            "temperature_c": {"type": "number"},
+            "advisory": {"type": "string"},
+        },
+        "required": ["location", "conditions", "temperature_c", "advisory"],
+        "additionalProperties": False,
+    }
+
+    async with (
+        temporary_chat_client(agent_name="RuntimeSchemaAgent") as chat_client,
+        ChatAgent(chat_client=chat_client) as agent,
+    ):
+        response = await agent.run(
+            "Give a brief weather digest for Seattle.",
+            additional_chat_options={
+                "response_format": {
+                    "type": "json_schema",
+                    "json_schema": {
+                        "name": runtime_schema["title"],
+                        "strict": True,
+                        "schema": runtime_schema,
+                    },
+                },
+            },
+        )
+
+        # Validate response
+        assert isinstance(response, AgentRunResponse)
+        assert response.text is not None
+
+        # Parse JSON and validate structure
+        import json
+
+        parsed = json.loads(response.text)
+        assert "location" in parsed
+        assert "conditions" in parsed
+        assert "temperature_c" in parsed
+        assert "advisory" in parsed
diff --git a/python/packages/core/agent_framework/openai/_responses_client.py b/python/packages/core/agent_framework/openai/_responses_client.py
index 9e50677fae..579452ef62 100644
--- a/python/packages/core/agent_framework/openai/_responses_client.py
+++ b/python/packages/core/agent_framework/openai/_responses_client.py
@@ -439,16 +439,23 @@ async def _prepare_options(
         if (tool_choice := run_options.get("tool_choice")) and isinstance(tool_choice, dict) and "mode" in tool_choice:
             run_options["tool_choice"] = tool_choice["mode"]
 
-        # additional properties
+        # additional properties (excluding response_format which is handled separately)
         additional_options = {
-            key: value for key, value in chat_options.additional_properties.items() if value is not None
+            key: value
+            for key, value in chat_options.additional_properties.items()
+            if value is not None and key != "response_format"
         }
         if additional_options:
             run_options.update(additional_options)
 
         # response format and text config (after additional_properties so user can pass text via additional_properties)
-        response_format = chat_options.response_format
-        text_config = run_options.pop("text", None)
+        # Check both chat_options.response_format and additional_properties for response_format
+        response_format: Any = (
+            chat_options.response_format
+            if chat_options.response_format is not None
+            else chat_options.additional_properties.get("response_format")
+        )
+        text_config: Any = run_options.pop("text", None)
         response_format, text_config = self._prepare_response_and_text_format(
             response_format=response_format, text_config=text_config
         )
diff --git a/python/packages/core/tests/openai/test_openai_responses_client.py b/python/packages/core/tests/openai/test_openai_responses_client.py
index a3c7ff5323..03510a2345 100644
--- a/python/packages/core/tests/openai/test_openai_responses_client.py
+++ b/python/packages/core/tests/openai/test_openai_responses_client.py
@@ -2355,3 +2355,91 @@ async def test_openai_responses_client_agent_local_mcp_tool() -> None:
         assert len(response.text) > 0
         # Should contain Azure-related content since it's asking about Azure CLI
         assert any(term in response.text.lower() for term in ["azure", "storage", "account", "cli"])
+
+
+class ReleaseBrief(BaseModel):
+    """Structured output model for release brief testing."""
+
+    title: str
+    summary: str
+    highlights: list[str]
+    model_config = {"extra": "forbid"}
+
+
+@pytest.mark.flaky
+@skip_if_openai_integration_tests_disabled
+async def test_openai_responses_client_agent_with_response_format_pydantic() -> None:
+    """Integration test for response_format with Pydantic model using OpenAI Responses Client."""
+    async with ChatAgent(
+        chat_client=OpenAIResponsesClient(),
+        instructions="You are a helpful assistant that returns structured JSON responses.",
+    ) as agent:
+        response = await agent.run(
+            "Summarize the following release notes into a ReleaseBrief:\n\n"
+            "Version 2.0 Release Notes:\n"
+            "- Added new streaming API for real-time responses\n"
+            "- Improved error handling with detailed messages\n"
+            "- Performance boost of 50% in batch processing\n"
+            "- Fixed memory leak in connection pooling",
+            response_format=ReleaseBrief,
+        )
+
+        # Validate response
+        assert isinstance(response, AgentRunResponse)
+        assert response.value is not None
+        assert isinstance(response.value, ReleaseBrief)
+
+        # Validate structured output fields
+        brief = response.value
+        assert len(brief.title) > 0
+        assert len(brief.summary) > 0
+        assert len(brief.highlights) > 0
+
+
+@pytest.mark.flaky
+@skip_if_openai_integration_tests_disabled
+async def test_openai_responses_client_agent_with_runtime_json_schema() -> None:
+    """Integration test for response_format with runtime JSON schema using OpenAI Responses Client."""
+    runtime_schema = {
+        "title": "WeatherDigest",
+        "type": "object",
+        "properties": {
+            "location": {"type": "string"},
+            "conditions": {"type": "string"},
+            "temperature_c": {"type": "number"},
+            "advisory": {"type": "string"},
+        },
+        "required": ["location", "conditions", "temperature_c", "advisory"],
+        "additionalProperties": False,
+    }
+
+    async with ChatAgent(
+        chat_client=OpenAIResponsesClient(),
+        instructions="Return only JSON that matches the provided schema. Do not add commentary.",
+    ) as agent:
+        response = await agent.run(
+            "Give a brief weather digest for Seattle.",
+            additional_chat_options={
+                "response_format": {
+                    "type": "json_schema",
+                    "json_schema": {
+                        "name": runtime_schema["title"],
+                        "strict": True,
+                        "schema": runtime_schema,
+                    },
+                },
+            },
+        )
+
+        # Validate response
+        assert isinstance(response, AgentRunResponse)
+        assert response.text is not None
+
+        # Parse JSON and validate structure
+        import json
+
+        parsed = json.loads(response.text)
+        assert "location" in parsed
+        assert "conditions" in parsed
+        assert "temperature_c" in parsed
+        assert "advisory" in parsed