Support for newer transcribe models such as gpt-4o-mini-transcribe-2025-12-15 #2200
Description
Describe the feature
OpenAI released new models for realtime and transcription, but the openai agents realtime models are not supporting them.
When will these models supported? For example when gpt-4o-mini-transcribe-2025-12-15 is used in the latest openai agents package(0.6.3);
File "/venv/lib64/python3.11/site-packages/agents/realtime/session.py", line 148, in __aenter__
await self._model.connect(model_config)
File "/venv/lib64/python3.11/site-packages/agents/realtime/openai_realtime.py", line 254, in connect
await self._update_session_config(model_settings)
File "/venv/lib64/python3.11/site-packages/agents/realtime/openai_realtime.py", line 823, in _update_session_config
session_config = self._get_session_config(model_settings)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/venv/lib64/python3.11/site-packages/agents/realtime/openai_realtime.py", line 886, in _get_session_config
input=OpenAIRealtimeAudioInput(**audio_input_args), # type: ignore[arg-type]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/venv/lib64/python3.11/site-packages/pydantic/main.py", line 250, in __init__
validated_self = self.__pydantic_validator__.validate_python(data, self_instance=self)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
pydantic_core._pydantic_core.ValidationError: 1 validation error for RealtimeAudioConfigInput
transcription.model
Input should be 'whisper-1', 'gpt-4o-mini-transcribe', 'gpt-4o-transcribe' or 'gpt-4o-transcribe-diarize' [type=literal_error, input_value='gpt-4o-mini-transcribe-2025-12-15', input_type=str]
For further information visit https://errors.pydantic.dev/2.12/v/literal_error