microsoft/autogen · error · ValueError

json_output must be a boolean or a Pydantic model class, got

Error message

json_output must be a boolean or a Pydantic model class, got {type(json_output)}

What it means

The json_output parameter of OllamaChatCompletionClient.create() must be None, a bool, or a Pydantic BaseModel subclass. The final else in the elif chain raises ValueError echoing the received type. Like error 803, passing a BaseModel *instance* (not the class) is the most frequent trigger, along with strings from config files.

Source

Thrown at python/packages/autogen-ext/src/autogen_ext/models/ollama/_ollama_client.py:560

        if json_output is not None:
            if self.model_info["json_output"] is False and json_output is True:
                raise ValueError("Model does not support JSON output.")
            if json_output is True:
                # JSON mode.
                response_format_value = "json"
            elif json_output is False:
                # Text mode.
                response_format_value = None
            elif isinstance(json_output, type) and issubclass(json_output, BaseModel):
                if response_format_value is not None:
                    raise ValueError(
                        "response_format and json_output cannot be set to a Pydantic model class at the same time. "
                        "Use json_output instead."
                    )
                # Beta client mode with Pydantic model class.
                response_format_value = json_output.model_json_schema()
            else:
                raise ValueError(f"json_output must be a boolean or a Pydantic model class, got {type(json_output)}")

        if "format" in create_args:
            # Handle the case where format is set from create_args.
            if json_output is not None:
                raise ValueError("json_output and format cannot be set at the same time. Use json_output instead.")
            assert response_format_value is None
            response_format_value = create_args["format"]
            # Remove format from create_args to prevent passing it twice.
            del create_args["format"]

        # TODO: allow custom handling.
        # For now we raise an error if images are present and vision is not supported
        if self.model_info["vision"] is False:
            for message in messages:
                if isinstance(message, UserMessage):
                    if isinstance(message.content, list) and any(isinstance(x, Image) for x in message.content):
                        raise ValueError("Model does not support vision and image was provided")

View on GitHub (pinned to 027ecf0a37)

Solutions

1. Pass the class: json_output=MyModel
2. Coerce config values before the call: parse 'true'/'false' strings to bool
3. Validate at the boundary: accept only bool | None | type[BaseModel] in your own wrapper's signature

Example fix

# before
await client.create(messages, json_output=SummaryResult())  # instance

# after
await client.create(messages, json_output=SummaryResult)  # class

def normalize_json_output(v: object) -> bool | type[BaseModel] | None:
    if v is None or isinstance(v, bool):
        return v
    if isinstance(v, type) and issubclass(v, BaseModel):
        return v
    if isinstance(v, str) and v.lower() in ("true", "false"):
        return v.lower() == "true"
    raise TypeError(f"unusable json_output: {v!r}")

result = await client.create(messages, json_output=normalize_json_output(cfg.get("json_output")))

def is_valid_ollama_json_output(v: object) -> TypeGuard[bool | type[BaseModel] | None]:
    return v is None or isinstance(v, bool) or (isinstance(v, type) and issubclass(v, BaseModel))

try:
    result = await client.create(messages, json_output=jo)
except ValueError as e:
    if "json_output must be a boolean or a Pydantic model class" in str(e):
        result = await client.create(messages, json_output=None)
    else:
        raise

When it happens

Trigger: json_output=MyModel() (instance); json_output='True' or 'true' from YAML/JSON config; json_output=1; forwarding a TypedDict or dataclass class.

Common situations: Deserialized configuration where booleans arrive as strings; passing a pre-built schema instance from another module; generic wrapper code that forwards arbitrary kwargs into json_output.

Related errors

• json_output must be a boolean, a BaseModel subclass or None.
• response_format must be a Pydantic model class, not {type(va
• The from field must be null or the agent name
• Invalid Role
• Unknown content type: {part}

AI-assisted analysis of microsoft/autogen@027ecf0a37 (2026-08-15). Data as JSON: /api/errors/61bb9cf55ed8c816. Report an issue: GitHub.