langchain-ai/langchain · error · OutputParserException

This output parser can only be used with a chat generation.

Error message

This output parser can only be used with a chat generation.

What it means

`JsonOpenAIFunctionCaller`/`OpenAIFunctionCallerOutputParser.parse_result` requires the first generation to be a `ChatGeneration` (a generation carrying a BaseMessage), because it reads `message.additional_kwargs["function_call"]`. Passing it a plain `Generation` (from a completion-style LLM rather than a chat model) fails immediately with OutputParserException. It exists to fail fast rather than AttributeError deeper in.

Source

Thrown at libs/core/langchain_core/output_parsers/openai_functions.py:45

    @override
    def parse_result(self, result: list[Generation], *, partial: bool = False) -> Any:
        """Parse the result of an LLM call to a JSON object.

        Args:
            result: The result of the LLM call.
            partial: Whether to parse partial JSON objects.

        Returns:
            The parsed JSON object.

        Raises:
            OutputParserException: If the output is not valid JSON.
        """
        generation = result[0]
        if not isinstance(generation, ChatGeneration):
            msg = "This output parser can only be used with a chat generation."
            raise OutputParserException(msg)
        message = generation.message
        try:
            func_call = copy.deepcopy(message.additional_kwargs["function_call"])
        except KeyError as exc:
            msg = f"Could not parse function call: {exc}"
            raise OutputParserException(msg) from exc

        if self.args_only:
            return func_call["arguments"]
        return func_call


class JsonOutputFunctionsParser(BaseCumulativeTransformOutputParser[Any]):
    """Parse an output as the JSON object."""

    strict: bool = False
    """Whether to allow non-JSON-compliant strings.

View on GitHub (pinned to e32fa9a52e)

Solutions

1. Use a chat model (`ChatOpenAI` or other `BaseChatModel`) in the chain so generations are `ChatGeneration`s.
2. If you must parse a completion-style function call, extract the JSON yourself from `result[0].text` with a JSON parser.
3. Prefer the modern tool-calling path (`llm.bind_tools`, `with_structured_output`) over the deprecated `function_call` kwargs API.

Example fix

// before
from langchain_openai import OpenAI
chain = prompt | OpenAI() | openai_functions.OpenAIFunctionCallerOutputParser()

// after
from langchain_openai import ChatOpenAI
chain = prompt | ChatOpenAI(model="gpt-4o") | openai_functions.OpenAIFunctionCallerOutputParser()

from langchain_core.outputs import ChatGeneration

def is_chat_generation(result) -> bool:
    return bool(result) and isinstance(result[0], ChatGeneration)

if not is_chat_generation(result):
    raise ValueError("use a chat model with this parser")

from langchain_core.outputs import ChatGeneration

def is_chat_gen(g) -> bool:
    return isinstance(g, ChatGeneration)

from langchain_core.exceptions import OutputParserException

try:
    parsed = parser.parse_result(result)
except OutputParserException as e:
    if "chat generation" in str(e):
        # switch chain to a BaseChatModel, or parse completion text manually
        ...

When it happens

Trigger: Piping a completion-style LLM (`OpenAI` legacy, any `LLM` subclass returning `Generation`) into a chain ending in this parser instead of a chat model; manually calling `parse_result([Generation(text=...)])`.

Common situations: Older tutorials built around `OpenAI` completion API and `create_openai_fn_chain`; migrating chains from completions to chat models piecemeal; unit tests constructing bare Generations.

Related errors

• Could not parse function call: {exc}
• Cannot concatenate FunctionMessageChunks with different name
• Trying to deserialize something that cannot be deserialized
• mime_type key is required for base64 data.
• Unsupported source type. Only 'url' and 'base64' are support

AI-assisted analysis of langchain-ai/langchain@e32fa9a52e (2026-08-14). Data as JSON: /api/errors/3d8c6234d0032864. Report an issue: GitHub.