diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
index 171d36f716b..8aba00e1c21 100644
--- a/.github/workflows/test.yml
+++ b/.github/workflows/test.yml
@@ -252,6 +252,25 @@ jobs:
       - name: Run tests
         run: tox -e py314-test-instrumentation-openai-v2-latest -- -ra
 
+  py313-test-instrumentation-openai-v2-pydantic1_ubuntu-latest:
+    name: instrumentation-openai-v2-pydantic1 3.13 Ubuntu
+    runs-on: ubuntu-latest
+    timeout-minutes: 30
+    steps:
+      - name: Checkout repo @ SHA - ${{ github.sha }}
+        uses: actions/checkout@v4
+
+      - name: Set up Python 3.13
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.13"
+
+      - name: Install tox
+        run: pip install tox-uv
+
+      - name: Run tests
+        run: tox -e py313-test-instrumentation-openai-v2-pydantic1 -- -ra
+
   pypy3-test-instrumentation-openai-v2-oldest_ubuntu-latest:
     name: instrumentation-openai-v2-oldest pypy-3.9 Ubuntu
     runs-on: ubuntu-latest
diff --git a/instrumentation-genai/opentelemetry-instrumentation-openai-v2/CHANGELOG.md b/instrumentation-genai/opentelemetry-instrumentation-openai-v2/CHANGELOG.md
index 76d2deabc45..6fc545df24c 100644
--- a/instrumentation-genai/opentelemetry-instrumentation-openai-v2/CHANGELOG.md
+++ b/instrumentation-genai/opentelemetry-instrumentation-openai-v2/CHANGELOG.md
@@ -19,6 +19,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   ([#4280](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/4280))
 - Add async wrappers for OpenAI Responses API streams and response stream managers
   ([#4325](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/4325))
+- Add strongly typed Responses API extractors with validation and content
+  extraction improvements
+  ([#4337](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/4337))
 
 ## Version 2.3b0 (2025-12-24)
 
diff --git a/instrumentation-genai/opentelemetry-instrumentation-openai-v2/src/opentelemetry/instrumentation/openai_v2/response_extractors.py b/instrumentation-genai/opentelemetry-instrumentation-openai-v2/src/opentelemetry/instrumentation/openai_v2/response_extractors.py
new file mode 100644
index 00000000000..2d5e8702b19
--- /dev/null
+++ b/instrumentation-genai/opentelemetry-instrumentation-openai-v2/src/opentelemetry/instrumentation/openai_v2/response_extractors.py
@@ -0,0 +1,419 @@
+# Copyright The OpenTelemetry Authors
+#
+# 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.
+
+from __future__ import annotations
+
+import logging
+from collections.abc import Mapping, Sequence
+from typing import TYPE_CHECKING, List, Optional, TypeVar, Union
+
+from pydantic import BaseModel, Field, StrictInt, StrictStr, ValidationError
+
+from opentelemetry.semconv._incubating.attributes import (
+    openai_attributes as OpenAIAttributes,
+)
+
+_PYDANTIC_V2 = hasattr(BaseModel, "model_validate")
+
+if _PYDANTIC_V2:
+    from pydantic import ConfigDict
+else:
+    ConfigDict = None
+
+if TYPE_CHECKING:
+    from opentelemetry.util.genai.types import (
+        InputMessage,
+        LLMInvocation,
+        OutputMessage,
+        Text,
+    )
+
+try:
+    from opentelemetry.util.genai.types import (
+        InputMessage,
+        OutputMessage,
+        Text,
+    )
+except ImportError:
+    InputMessage = None
+    OutputMessage = None
+    Text = None
+
+GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS = "gen_ai.usage.cache_read.input_tokens"
+GEN_AI_USAGE_CACHE_CREATION_INPUT_TOKENS = (
+    "gen_ai.usage.cache_creation.input_tokens"
+)
+
+_logger = logging.getLogger(__name__)
+ModelT = TypeVar("ModelT", bound=BaseModel)
+
+
+class _ExtractorModel(BaseModel):
+    if _PYDANTIC_V2:
+        model_config = ConfigDict(extra="ignore", from_attributes=True)
+    else:
+
+        class Config:
+            extra = "ignore"
+            orm_mode = True
+
+
+class _ResponseTextFormatModel(_ExtractorModel):
+    type: Optional[StrictStr] = None
+
+
+class _ResponseTextConfigModel(_ExtractorModel):
+    format: Optional[_ResponseTextFormatModel] = None
+
+
+class _ResponseInputContentModel(_ExtractorModel):
+    text: Optional[StrictStr] = None
+
+
+class _ResponseInputItemModel(_ExtractorModel):
+    role: Optional[StrictStr] = None
+    content: Optional[Union[StrictStr, List[_ResponseInputContentModel]]] = (
+        None
+    )
+
+
+class _ResponsesRequestModel(_ExtractorModel):
+    instructions: Optional[StrictStr] = None
+    input: Optional[Union[StrictStr, List[_ResponseInputItemModel]]] = None
+    text: Optional[_ResponseTextConfigModel] = None
+
+
+class _ResponseOutputContentModel(_ExtractorModel):
+    type: Optional[StrictStr] = None
+    text: Optional[StrictStr] = None
+    refusal: Optional[StrictStr] = None
+
+
+class _ResponseOutputItemModel(_ExtractorModel):
+    type: Optional[StrictStr] = None
+    role: Optional[StrictStr] = None
+    status: Optional[StrictStr] = None
+    content: List[_ResponseOutputContentModel] = Field(default_factory=list)
+
+
+class _UsageDetailsModel(_ExtractorModel):
+    cached_tokens: Optional[StrictInt] = None
+    cache_creation_input_tokens: Optional[StrictInt] = None
+
+
+class _UsageModel(_ExtractorModel):
+    input_tokens: Optional[StrictInt] = None
+    output_tokens: Optional[StrictInt] = None
+    prompt_tokens: Optional[StrictInt] = None
+    completion_tokens: Optional[StrictInt] = None
+    input_tokens_details: Optional[_UsageDetailsModel] = None
+    prompt_tokens_details: Optional[_UsageDetailsModel] = None
+
+
+class _ResponsesResultModel(_ExtractorModel):
+    # Responses can be partial or otherwise omit `output`. Treat that as "no
+    # terminal message items available yet" so downstream extraction helpers
+    # naturally return empty lists instead of raising.
+    output: List[_ResponseOutputItemModel] = Field(default_factory=list)
+    model: Optional[StrictStr] = None
+    id: Optional[StrictStr] = None
+    service_tier: Optional[StrictStr] = None
+    usage: Optional[_UsageModel] = None
+
+
+def _rebuild_model(model_type: type[BaseModel]) -> None:
+    if _PYDANTIC_V2:
+        model_type.model_rebuild(_types_namespace=globals())
+    else:
+        model_type.update_forward_refs(**globals())
+
+
+for _model_type in (
+    _ResponseTextFormatModel,
+    _ResponseTextConfigModel,
+    _ResponseInputContentModel,
+    _ResponseInputItemModel,
+    _ResponsesRequestModel,
+    _ResponseOutputContentModel,
+    _ResponseOutputItemModel,
+    _UsageDetailsModel,
+    _UsageModel,
+    _ResponsesResultModel,
+):
+    _rebuild_model(_model_type)
+
+
+def _validate_model(
+    model_type: type[ModelT], value: object, context: str
+) -> ModelT | None:
+    try:
+        if _PYDANTIC_V2:
+            return model_type.model_validate(value)
+        if isinstance(value, Mapping):
+            return model_type.parse_obj(value)
+        return model_type.from_orm(value)
+    except ValidationError:
+        _logger.debug(
+            "OpenAI responses extractor validation failed for %s",
+            context,
+            exc_info=True,
+        )
+        return None
+
+
+def _validate_request_kwargs(
+    kwargs: Mapping[str, object],
+) -> _ResponsesRequestModel | None:
+    return _validate_model(
+        _ResponsesRequestModel, kwargs, "request parameters"
+    )
+
+
+def _validate_response_result(result: object) -> _ResponsesResultModel | None:
+    return _validate_model(_ResponsesResultModel, result, "response payload")
+
+
+def _coerce_response_result(
+    result: object | _ResponsesResultModel | None,
+) -> _ResponsesResultModel | None:
+    if result is None:
+        return None
+
+    if isinstance(result, _ResponsesResultModel):
+        return result
+
+    return _validate_response_result(result)
+
+
+def _extract_system_instruction(kwargs: Mapping[str, object]) -> list["Text"]:
+    """Extract system instruction from the ``instructions`` parameter."""
+    if Text is None:
+        return []
+
+    request = _validate_request_kwargs(kwargs)
+    if request is None or request.instructions is None:
+        return []
+
+    return [Text(content=request.instructions)]
+
+
+def _extract_input_messages(
+    kwargs: Mapping[str, object],
+) -> list["InputMessage"]:
+    """Extract input messages from Responses API kwargs."""
+    if InputMessage is None or Text is None:
+        return []
+
+    request = _validate_request_kwargs(kwargs)
+    if request is None or request.input is None:
+        return []
+
+    if isinstance(request.input, str):
+        return [InputMessage(role="user", parts=[Text(content=request.input)])]
+
+    messages: list[InputMessage] = []
+    for item in request.input:
+        if item.role is None:
+            continue
+
+        if isinstance(item.content, str):
+            messages.append(
+                InputMessage(
+                    role=item.role, parts=[Text(content=item.content)]
+                )
+            )
+            continue
+
+        if item.content is None:
+            continue
+
+        parts = [
+            Text(content=part.text)
+            for part in item.content
+            if part.text is not None
+        ]
+        if parts:
+            messages.append(InputMessage(role=item.role, parts=parts))
+
+    return messages
+
+
+def _extract_output_parts(
+    content_blocks: Sequence[_ResponseOutputContentModel],
+) -> list["Text"]:
+    if Text is None:
+        return []
+
+    parts: list[Text] = []
+    for block in content_blocks:
+        if block.type == "output_text" and block.text is not None:
+            parts.append(Text(content=block.text))
+        elif block.type == "refusal" and block.refusal is not None:
+            parts.append(Text(content=block.refusal))
+    return parts
+
+
+def _finish_reason_from_status(status: str | None) -> str | None:
+    # Responses API output items expose lifecycle statuses rather than finish
+    # reasons. We map the normal terminal state to the GenAI "stop" reason,
+    # preserve the other terminal statuses verbatim, and drop non-terminal or
+    # unknown states from the finish_reasons attribute.
+    if status == "completed":
+        return "stop"
+    if status in {"failed", "cancelled", "incomplete"}:
+        return status
+    return None
+
+
+def _extract_output_messages_from_model(
+    result: _ResponsesResultModel,
+) -> list["OutputMessage"]:
+    if OutputMessage is None or Text is None:
+        return []
+
+    messages: list[OutputMessage] = []
+    for item in result.output:
+        if item.type != "message":
+            continue
+        finish_reason = _finish_reason_from_status(item.status)
+        if finish_reason is None:
+            continue
+
+        messages.append(
+            OutputMessage(
+                role=item.role if item.role is not None else "assistant",
+                parts=_extract_output_parts(item.content),
+                finish_reason=finish_reason,
+            )
+        )
+
+    return messages
+
+
+def _extract_output_messages(
+    result: object | _ResponsesResultModel | None,
+) -> list["OutputMessage"]:
+    """Extract output messages from a Responses API result."""
+    validated_result = _coerce_response_result(result)
+    if validated_result is None:
+        return []
+
+    return _extract_output_messages_from_model(validated_result)
+
+
+def _extract_finish_reasons_from_model(
+    result: _ResponsesResultModel,
+) -> list[str]:
+    finish_reasons: list[str] = []
+    for item in result.output:
+        if item.type != "message":
+            continue
+        finish_reason = _finish_reason_from_status(item.status)
+        if finish_reason is not None:
+            finish_reasons.append(finish_reason)
+    return finish_reasons
+
+
+def _extract_finish_reasons(
+    result: object | _ResponsesResultModel | None,
+) -> list[str]:
+    """Extract finish reasons from Responses API output items."""
+    validated_result = _coerce_response_result(result)
+    if validated_result is None:
+        return []
+
+    return _extract_finish_reasons_from_model(validated_result)
+
+
+def _extract_output_type(kwargs: Mapping[str, object]) -> str | None:
+    """Extract output type from Responses API request text.format."""
+    request = _validate_request_kwargs(kwargs)
+    if request is None or request.text is None or request.text.format is None:
+        return None
+
+    if request.text.format.type == "json_schema":
+        return "json"
+    return request.text.format.type
+
+
+def _set_invocation_usage_attributes(
+    invocation: "LLMInvocation", usage: _UsageModel
+) -> None:
+    if usage.input_tokens is not None:
+        invocation.input_tokens = usage.input_tokens
+    else:
+        invocation.input_tokens = usage.prompt_tokens
+
+    if usage.output_tokens is not None:
+        invocation.output_tokens = usage.output_tokens
+    else:
+        invocation.output_tokens = usage.completion_tokens
+
+    details = (
+        usage.input_tokens_details
+        if usage.input_tokens_details is not None
+        else usage.prompt_tokens_details
+    )
+    if details is None:
+        return
+
+    if details.cached_tokens is not None:
+        invocation.attributes[GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS] = (
+            details.cached_tokens
+        )
+
+    if details.cache_creation_input_tokens is not None:
+        invocation.attributes[GEN_AI_USAGE_CACHE_CREATION_INPUT_TOKENS] = (
+            details.cache_creation_input_tokens
+        )
+
+
+def _set_invocation_response_attributes(
+    invocation: "LLMInvocation",
+    result: object | None,
+    capture_content: bool,
+) -> None:
+    # This helper sits at the raw response boundary for instrumentation. Unlike
+    # the extractor helpers above, it intentionally validates exactly once and
+    # then fans out to the `_from_model` helpers rather than accepting an
+    # already-coerced `_ResponsesResultModel`.
+    if result is None:
+        return
+
+    validated_result = _validate_response_result(result)
+    if validated_result is None:
+        return
+
+    if validated_result.model is not None:
+        invocation.response_model_name = validated_result.model
+
+    if validated_result.id is not None:
+        invocation.response_id = validated_result.id
+
+    if validated_result.service_tier is not None:
+        invocation.attributes[
+            OpenAIAttributes.OPENAI_RESPONSE_SERVICE_TIER
+        ] = validated_result.service_tier
+
+    if validated_result.usage is not None:
+        _set_invocation_usage_attributes(invocation, validated_result.usage)
+
+    finish_reasons = _extract_finish_reasons_from_model(validated_result)
+    if finish_reasons:
+        invocation.finish_reasons = finish_reasons
+
+    if capture_content:
+        output_messages = _extract_output_messages_from_model(validated_result)
+        if output_messages:
+            invocation.output_messages = output_messages
diff --git a/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/requirements.pydantic1.txt b/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/requirements.pydantic1.txt
new file mode 100644
index 00000000000..9ca166184dc
--- /dev/null
+++ b/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/requirements.pydantic1.txt
@@ -0,0 +1,55 @@
+# Copyright The OpenTelemetry Authors
+#
+# 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.
+
+
+# ********************************
+# WARNING: NOT HERMETIC !!!!!!!!!!
+# ********************************
+#
+# This "requirements.txt" is installed in conjunction
+# with multiple other dependencies in the top-level "tox.ini"
+# file. In particular, please see:
+#
+#   openai-pydantic1: {[testenv]test_deps}
+#   openai-pydantic1: -r {toxinidir}/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/requirements.pydantic1.txt
+#
+# This provides additional dependencies, namely:
+#
+#   opentelemetry-api
+#   opentelemetry-sdk
+#   opentelemetry-semantic-conventions
+#
+# ... with a "dev" version based on the latest distribution.
+
+
+# This variant of the requirements aims to test the system using
+# the newest supported version of external dependencies with Pydantic 1.x.
+
+openai==1.109.1
+pydantic==1.10.24
+httpx==0.27.2
+# older jiter is required for PyPy < 3.11
+jiter==0.11.1
+Deprecated==1.2.14
+importlib-metadata==6.11.0
+packaging==24.0
+pytest==7.4.4
+pytest-vcr==1.0.2
+pytest-asyncio==0.21.0
+wrapt==1.16.0
+# test with the latest version of opentelemetry-api, sdk, and semantic conventions
+
+-e opentelemetry-instrumentation
+-e instrumentation-genai/opentelemetry-instrumentation-openai-v2
+-e util/opentelemetry-util-genai
diff --git a/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/test_response_extractors.py b/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/test_response_extractors.py
new file mode 100644
index 00000000000..1cf9a1a70e1
--- /dev/null
+++ b/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/test_response_extractors.py
@@ -0,0 +1,365 @@
+# Copyright The OpenTelemetry Authors
+#
+# 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.
+
+import logging
+from types import SimpleNamespace
+from unittest import mock
+
+import pytest
+
+from opentelemetry.instrumentation.openai_v2 import response_extractors
+from opentelemetry.semconv._incubating.attributes import (
+    openai_attributes as OpenAIAttributes,
+)
+from opentelemetry.util.genai.types import LLMInvocation
+
+
+def _validate_compat_model(loaded_module, model_type, value):
+    return loaded_module._validate_model(model_type, value, "test")
+
+
+@pytest.fixture(scope="module", name="loaded_module")
+def _loaded_module_fixture():
+    return response_extractors
+
+
+@pytest.fixture(
+    scope="module", name="gen_ai_usage_cache_creation_input_tokens"
+)
+def _gen_ai_usage_cache_creation_input_tokens_fixture(loaded_module):
+    return loaded_module.GEN_AI_USAGE_CACHE_CREATION_INPUT_TOKENS
+
+
+@pytest.fixture(scope="module", name="gen_ai_usage_cache_read_input_tokens")
+def _gen_ai_usage_cache_read_input_tokens_fixture(loaded_module):
+    return loaded_module.GEN_AI_USAGE_CACHE_READ_INPUT_TOKENS
+
+
+def test_extract_system_instruction_returns_text_for_string(loaded_module):
+    instructions = loaded_module._extract_system_instruction(
+        {"instructions": "Be concise"}
+    )
+
+    assert [part.content for part in instructions] == ["Be concise"]
+
+
+def test_extract_input_messages_supports_string_and_mixed_message_content(
+    loaded_module,
+):
+    from_string = loaded_module._extract_input_messages({"input": "Hello"})
+    from_list = loaded_module._extract_input_messages(
+        {
+            "input": [
+                {"role": "user", "content": "First"},
+                SimpleNamespace(
+                    role="assistant",
+                    content=[
+                        {"text": "Second"},
+                        SimpleNamespace(text="Third"),
+                        {"type": "input_image", "image_url": "ignored"},
+                    ],
+                ),
+                {"role": None, "content": "ignored"},
+            ]
+        }
+    )
+
+    assert [
+        (msg.role, [part.content for part in msg.parts]) for msg in from_string
+    ] == [("user", ["Hello"])]
+    assert [
+        (msg.role, [part.content for part in msg.parts]) for msg in from_list
+    ] == [
+        ("user", ["First"]),
+        ("assistant", ["Second", "Third"]),
+    ]
+
+
+def test_extract_output_messages_maps_parts_and_finish_reasons(loaded_module):
+    result = SimpleNamespace(
+        output=[
+            SimpleNamespace(
+                type="message",
+                role="assistant",
+                status="completed",
+                content=[
+                    SimpleNamespace(type="output_text", text="Done"),
+                    SimpleNamespace(type="refusal", refusal="Cannot comply"),
+                    SimpleNamespace(type="summary", text="ignored"),
+                ],
+            ),
+            SimpleNamespace(
+                type="message",
+                status="incomplete",
+                content=[SimpleNamespace(type="output_text", text="Partial")],
+            ),
+            SimpleNamespace(
+                type="message",
+                status="queued",
+                content=[SimpleNamespace(type="output_text", text="Pending")],
+            ),
+            SimpleNamespace(type="tool_call", status="completed", content=[]),
+        ]
+    )
+
+    messages = loaded_module._extract_output_messages(result)
+
+    assert [(msg.role, msg.finish_reason) for msg in messages] == [
+        ("assistant", "stop"),
+        ("assistant", "incomplete"),
+    ]
+    assert [[part.content for part in msg.parts] for msg in messages] == [
+        ["Done", "Cannot comply"],
+        ["Partial"],
+    ]
+
+
+def test_extract_finish_reasons_only_reads_message_items(loaded_module):
+    result = SimpleNamespace(
+        output=[
+            SimpleNamespace(type="message", status="completed"),
+            SimpleNamespace(type="message", status=None),
+            SimpleNamespace(type="message", status="in_progress"),
+            SimpleNamespace(type="tool_call", status="incomplete"),
+        ]
+    )
+
+    assert loaded_module._extract_finish_reasons(result) == ["stop"]
+
+
+def test_extract_output_type_handles_text_format_mapping(loaded_module):
+    assert (
+        loaded_module._extract_output_type(
+            {"text": {"format": {"type": "json_schema"}}}
+        )
+        == "json"
+    )
+    assert (
+        loaded_module._extract_output_type(
+            {"text": {"format": {"type": "text"}}}
+        )
+        == "text"
+    )
+    assert (
+        loaded_module._extract_output_type(
+            {
+                "text": SimpleNamespace(
+                    format=SimpleNamespace(type="json_schema")
+                )
+            }
+        )
+        == "json"
+    )
+    assert (
+        loaded_module._extract_output_type(
+            {"text": SimpleNamespace(format=SimpleNamespace(type="text"))}
+        )
+        == "text"
+    )
+    # Invalid request shapes should degrade to no extracted output type rather
+    # than surfacing validation errors from instrumentation.
+    assert (
+        loaded_module._extract_output_type({"text": {"format": "plain"}})
+        is None
+    )
+    assert loaded_module._extract_output_type({"text": "plain"}) is None
+
+
+def test_extractors_handle_missing_genai_types_import(loaded_module):
+    with (
+        mock.patch.object(loaded_module, "Text", None),
+        mock.patch.object(loaded_module, "InputMessage", None),
+        mock.patch.object(loaded_module, "OutputMessage", None),
+    ):
+        assert (
+            loaded_module._extract_system_instruction({"instructions": "hi"})
+            == []
+        )
+        assert loaded_module._extract_input_messages({"input": "hi"}) == []
+        assert (
+            loaded_module._extract_output_messages(
+                SimpleNamespace(
+                    output=[SimpleNamespace(type="message", content=[])]
+                )
+            )
+            == []
+        )
+
+
+def test_set_invocation_response_attributes_populates_usage_and_metadata(
+    loaded_module,
+    gen_ai_usage_cache_creation_input_tokens,
+    gen_ai_usage_cache_read_input_tokens,
+):
+    invocation = LLMInvocation(request_model="gpt-4o-mini")
+    result = SimpleNamespace(
+        model="gpt-4.1",
+        id="resp_123",
+        service_tier="scale",
+        usage=SimpleNamespace(
+            prompt_tokens=11,
+            completion_tokens=7,
+            prompt_tokens_details=SimpleNamespace(
+                cached_tokens=3,
+                cache_creation_input_tokens=5,
+            ),
+        ),
+    )
+
+    loaded_module._set_invocation_response_attributes(
+        invocation, result, capture_content=False
+    )
+
+    assert invocation.response_model_name == "gpt-4.1"
+    assert invocation.response_id == "resp_123"
+    assert invocation.input_tokens == 11
+    assert invocation.output_tokens == 7
+    assert invocation.attributes == {
+        OpenAIAttributes.OPENAI_RESPONSE_SERVICE_TIER: "scale",
+        gen_ai_usage_cache_read_input_tokens: 3,
+        gen_ai_usage_cache_creation_input_tokens: 5,
+    }
+
+
+def test_set_invocation_response_attributes_accepts_mapping_usage(
+    loaded_module,
+    gen_ai_usage_cache_creation_input_tokens,
+    gen_ai_usage_cache_read_input_tokens,
+):
+    invocation = LLMInvocation(request_model="gpt-4o-mini")
+    result = SimpleNamespace(
+        usage={
+            "input_tokens": 13,
+            "output_tokens": 8,
+            "input_tokens_details": {
+                "cached_tokens": 2,
+                "cache_creation_input_tokens": 4,
+            },
+        }
+    )
+
+    loaded_module._set_invocation_response_attributes(
+        invocation, result, capture_content=False
+    )
+
+    assert invocation.input_tokens == 13
+    assert invocation.output_tokens == 8
+    assert invocation.attributes == {
+        gen_ai_usage_cache_read_input_tokens: 2,
+        gen_ai_usage_cache_creation_input_tokens: 4,
+    }
+
+
+def test_set_invocation_response_attributes_populates_output_messages(
+    loaded_module,
+):
+    invocation = LLMInvocation(request_model="gpt-4o-mini")
+    result = SimpleNamespace(
+        output=[
+            SimpleNamespace(
+                type="message",
+                role="assistant",
+                status="completed",
+                content=[SimpleNamespace(type="output_text", text="Done")],
+            )
+        ]
+    )
+
+    loaded_module._set_invocation_response_attributes(
+        invocation, result, capture_content=True
+    )
+
+    assert invocation.finish_reasons == ["stop"]
+    assert [
+        (message.role, message.finish_reason)
+        for message in invocation.output_messages
+    ] == [("assistant", "stop")]
+    assert [
+        [part.content for part in message.parts]
+        for message in invocation.output_messages
+    ] == [["Done"]]
+
+
+def test_prevalidated_response_model_skips_revalidation(
+    loaded_module, monkeypatch
+):
+    validated_result = _validate_compat_model(
+        loaded_module,
+        loaded_module._ResponsesResultModel,
+        SimpleNamespace(
+            output=[
+                SimpleNamespace(
+                    type="message",
+                    status="completed",
+                    content=[SimpleNamespace(type="output_text", text="Done")],
+                )
+            ]
+        ),
+    )
+    assert validated_result is not None
+
+    def _unexpected_validation(_result):
+        raise AssertionError("unexpected response revalidation")
+
+    monkeypatch.setattr(
+        loaded_module, "_validate_response_result", _unexpected_validation
+    )
+
+    assert loaded_module._extract_finish_reasons(validated_result) == ["stop"]
+    messages = loaded_module._extract_output_messages(validated_result)
+    assert [part.content for part in messages[0].parts] == ["Done"]
+
+
+@pytest.mark.parametrize(
+    ("kwargs", "extractor_name"),
+    [
+        ({"instructions": ["not-a-string"]}, "_extract_system_instruction"),
+        ({"input": 42}, "_extract_input_messages"),
+        ({"text": {"format": {"type": 42}}}, "_extract_output_type"),
+    ],
+)
+def test_request_validation_errors_are_logged_and_ignored(
+    loaded_module, caplog, kwargs, extractor_name
+):
+    caplog.set_level(logging.DEBUG, logger=loaded_module.__name__)
+    extractor = getattr(loaded_module, extractor_name)
+
+    result = extractor(kwargs)
+
+    assert result in ([], None)
+    assert "OpenAI responses extractor validation failed" in caplog.text
+
+
+def test_response_validation_errors_are_logged_and_ignored(
+    loaded_module, caplog
+):
+    caplog.set_level(logging.DEBUG, logger=loaded_module.__name__)
+    invocation = LLMInvocation(request_model="gpt-4o-mini")
+    invalid_result = SimpleNamespace(output=42, usage=42)
+
+    assert loaded_module._extract_output_messages(invalid_result) == []
+    assert loaded_module._extract_finish_reasons(invalid_result) == []
+
+    loaded_module._set_invocation_response_attributes(
+        invocation, invalid_result, capture_content=True
+    )
+
+    assert invocation.response_model_name is None
+    assert invocation.response_id is None
+    assert invocation.input_tokens is None
+    assert invocation.output_tokens is None
+    assert invocation.finish_reasons is None
+    assert not invocation.output_messages
+    assert not invocation.attributes
+    assert "OpenAI responses extractor validation failed" in caplog.text
diff --git a/tox.ini b/tox.ini
index b351bf13eee..5b13ed9365a 100644
--- a/tox.ini
+++ b/tox.ini
@@ -10,6 +10,7 @@ envlist =
 
     ; instrumentation-openai
     py3{9,10,11,12,13,14}-test-instrumentation-openai-v2-{oldest,latest}
+    py313-test-instrumentation-openai-v2-pydantic1
     pypy3-test-instrumentation-openai-v2-{oldest,latest}
     lint-instrumentation-openai-v2
 
@@ -482,6 +483,8 @@ deps =
   # and the latest version of OTel API and SDK
   openai-latest: {[testenv]test_deps}
   openai-latest: -r {toxinidir}/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/requirements.latest.txt
+  openai-pydantic1: {[testenv]test_deps}
+  openai-pydantic1: -r {toxinidir}/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/requirements.pydantic1.txt
   lint-instrumentation-openai-v2: -r {toxinidir}/instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/requirements.oldest.txt
   openai_agents-oldest: -r {toxinidir}/instrumentation-genai/opentelemetry-instrumentation-openai-agents-v2/tests/requirements.oldest.txt
   openai_agents-latest: {[testenv]test_deps}