`NativeOutput` for groq Agents using supported models

[Kukuksumusu]

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

Hi, according to the groq documentation, the "Structured Outputs" (or "JSON Schema Mode") should be supported, at least for some subset of models.
This should allow the user to use NativeOutput (as per https://ai.pydantic.dev/output/#native-output), but it's not the case.

Using the example from documentation with groq instead of openai results in pydantic_ai.exceptions.UserError: Native structured output is not supported by the model.

I'm not sure if this is a bug or rather feature request, but I marked it a bug, since I think that user would reasonably expect this to work, based on the documentation

Example Code

from pydantic_ai import Agent, NativeOutput
from pydantic import BaseModel


class Fruit(BaseModel):
    name: str
    color: str


class Vehicle(BaseModel):
    name: str
    wheels: int

agent = Agent(
    'groq:openai/gpt-oss-120b',
    output_type=NativeOutput(
        [Fruit, Vehicle],
        name='Fruit_or_vehicle',
        description='Return a fruit or vehicle.'
    ),
)
result = agent.run_sync('What is a Ford Explorer?')

This results in:

Traceback (most recent call last):
  ...
    result = agent.run_sync('What is a Ford Explorer?')
  File "...venv/lib/python3.13/site-packages/pydantic_ai/agent/abstract.py", line 306, in run_sync
    return get_event_loop().run_until_complete(
           ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^
        self.run(
        ^^^^^^^^^
    ...<11 lines>...
        )
        ^
    )
    ^
  File ".pyenv/versions/3.13.1/lib/python3.13/asyncio/base_events.py", line 720, in run_until_complete
    return future.result()
           ~~~~~~~~~~~~~^^
  File ".venv/lib/python3.13/site-packages/pydantic_ai/agent/abstract.py", line 200, in run
    async with self.iter(
               ~~~~~~~~~^
        user_prompt=user_prompt,
        ^^^^^^^^^^^^^^^^^^^^^^^^
    ...<7 lines>...
        toolsets=toolsets,
        ^^^^^^^^^^^^^^^^^^
    ) as agent_run:
    ^
  File ".pyenv/versions/3.13.1/lib/python3.13/contextlib.py", line 214, in __aenter__
    return await anext(self.gen)
           ^^^^^^^^^^^^^^^^^^^^^
  File ".venv/lib/python3.13/site-packages/pydantic_ai/agent/__init__.py", line 552, in iter
    output_schema = self._prepare_output_schema(output_type, model_used.profile)
  File ".venv/lib/python3.13/site-packages/pydantic_ai/agent/__init__.py", line 1313, in _prepare_output_schema
    schema.raise_if_unsupported(model_profile)
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^
  File ".venv/lib/python3.13/site-packages/pydantic_ai/_output.py", line 471, in raise_if_unsupported
    raise UserError('Native structured output is not supported by the model.')
pydantic_ai.exceptions.UserError: Native structured output is not supported by the model.

Python, Pydantic AI & LLM client version

Python | 3.13
pydantic-ai | 0.8.1
pydantic-ai-slim | 0.8.1

[DouweM]

@Kukuksumusu Thanks for the report, implemented in #2772!