Submission checklist

• This is a feature request, not a bug report or usage question.
• I added a clear and descriptive title that summarizes the feature request.
• I used the GitHub search to find a similar feature request and didn't find it.
• I checked the LangChain documentation and API reference to see if this feature already exists.
• This is not related to the langchain-community package.

Package (Required)

• langchain
• langchain-openai
• langchain-anthropic
• langchain-classic
• langchain-core
• langchain-model-profiles
• langchain-tests
• langchain-text-splitters
• langchain-chroma
• langchain-deepseek
• langchain-exa
• langchain-fireworks
• langchain-groq
• langchain-huggingface
• langchain-mistralai
• langchain-nomic
• langchain-ollama
• langchain-openrouter
• langchain-perplexity
• langchain-qdrant
• langchain-xai
• Other / not sure / general

Feature Description

Add support for OpenAI Responses API apply_patch in langchain-openai.

Expected behavior:

model = ChatOpenAI(model="gpt-5.2", output_version="responses/v1")
model_with_tools = model.bind_tools([
    {"type": "apply_patch"}
])
message = model_with_tools.invoke("Create hello.txt")
message should preserve OpenAI’s apply_patch_call output item instead of dropping it.

Use Case

I’m building a code-editing agent that uses OpenAI’s native apply_patch tool.

OpenAI returns this correctly:

{
  "type": "apply_patch_call",
  "call_id": "call_...",
  "operation": {
    "type": "create_file",
    "path": "hello.txt",
    "diff": "+hello\n"
  },
  "status": "completed"
}

But LangChain currently converts this into an AIMessage with:

content = []
tool_calls = []
additional_kwargs = {}

So the patch operation is lost. This also affects streaming, where apply_patch_call events are not surfaced in chunks or the final accumulated message.

Proposed Solution

I think this could be implemented by:

• Adding apply_patch to the OpenAI well-known tool lists so bind_tools([{"type": "apply_patch"}]) works.
• Preserving apply_patch_call and apply_patch_call_output in langchain_openai.chat_models.base._construct_lc_result_from_responses_api.
• Preserving the same item types in _convert_responses_chunk_to_generation_chunk for streaming.
• Optionally supporting round-trip conversion when an AIMessage containing these Responses content blocks is sent back as input.

The API could look like:

model = ChatOpenAI(model="gpt-5.2", output_version="responses/v1")
model_with_tools = model.bind_tools([
    {"type": "apply_patch"}
])
message = model_with_tools.invoke("Create hello.txt")
message.content
# [
#   {
#     "type": "apply_patch_call",
#     "call_id": "call_...",
#     "operation": {
#       "type": "create_file",
#       "path": "hello.txt",
#       "diff": "+hello\n",
#     },
#     "status": "completed",
#   }
# ]

Alternatives Considered

No response

Additional Context

I tried using raw .bind(...):

model.bind(
    tools=[{"type": "apply_patch"}],
    tool_choice={"type": "apply_patch"},
)

This sends the tool to OpenAI correctly, but LangChain still drops the returned apply_patch_call.

I also tested a local monkeypatch that adds apply_patch_call to the Responses conversion allow-list. That works, but it relies on private LangChain internals.