supports_json_schema_output and some other profile settings should be provider settings

[xcpky]

Initial Checks

• I confirm that I'm using the latest version of Pydantic AI
• I confirm that I searched for my issue in https://github.com/pydantic/pydantic-ai/issues before opening this issue

Description

supports_json_schema_output: bool = False
"""Whether the model supports JSON schema output."""
supports_json_object_output: bool = False
"""Whether the model supports JSON object output."""
default_structured_output_mode: StructuredOutputMode = 'tool'
"""The default structured output mode to use for the model."""
prompted_output_template: str = dedent(
    """
    Always respond with a JSON object that's compatible with this schema:

    {schema}

    Don't include any text or Markdown fencing before or after.
    """
)
"""The instructions template to use for prompted structured output. The '{schema}' placeholder will be replaced with the JSON schema for the output."""
json_schema_transformer: type[JsonSchemaTransformer] | None = None
"""The transformer to use to make JSON schemas for tools and structured output compatible with the model."""

ModelProfile Describes how requests to and responses from specific models or families of models need to be constructed and processed to get the best results, independent of the model and provider classes used.

The above fields of ModelProfile are not model-specific, but provider-specific. They should be categorized as provider settings. For example, there are a bunch of providers for deepseek v3.2, which have different support for json schema and json object.

Example Code

Python, Pydantic AI & LLM client version

Python 3.13
pydantic_ai 1.12.0

[DouweM]

The above fields of ModelProfile are not model-specific, but provider-specific.

@xcpky They're really specific to the combination of model and provider as a provider that supports JSON schema with one model does not necessarily support it for all models. For example:

pydantic-ai/pydantic_ai_slim/pydantic_ai/profiles/google.py

Lines 11 to 20 in 86b645f

	def google_model_profile(model_name: str) -> ModelProfile | None:
	"""Get the model profile for a Google model."""
	is_image_model = 'image' in model_name
	return ModelProfile(
	json_schema_transformer=GoogleJsonSchemaTransformer,
	supports_image_output=is_image_model,
	supports_json_schema_output=not is_image_model,
	supports_json_object_output=not is_image_model,
	supports_tools=not is_image_model,
	)

For that reason, Provider subclasses can already implement the model_profile method to override what's defined on the model's own profile.

Does that not give you enough flexibility to represent what you're looking for?