[8/n] Make realtime more like the rest of agents sdk

Key changes:
1. Transport -> model.
2. Extract any model settings into `RealtimeSessionModelSettings`.
3. RealtimeRunConfig, similar to the RunConfig in `run.py`.

Next PR I'll update session to be better.

Author: rm-openai

src/agents/realtime/__init__.py

@@ -1,5 +1,16 @@
 from .agent import RealtimeAgent, RealtimeAgentHooks, RealtimeRunHooks
-from .config import APIKeyOrKeyFunc
+from .config import (
+    RealtimeAudioFormat,
+    RealtimeClientMessage,
+    RealtimeInputAudioTranscriptionConfig,
+    RealtimeModelName,
+    RealtimeRunConfig,
+    RealtimeSessionModelSettings,
+    RealtimeTurnDetectionConfig,
+    RealtimeUserInput,
+    RealtimeUserInputMessage,
+    RealtimeUserInputText,
+)
 from .events import (
     RealtimeAgentEndEvent,
     RealtimeAgentStartEvent,
@@ -15,13 +26,12 @@
     RealtimeToolEnd,
     RealtimeToolStart,
 )
-from .session import RealtimeSession
-from .transport import (
-    RealtimeModelName,
-    RealtimeSessionTransport,
-    RealtimeTransportConnectionOptions,
-    RealtimeTransportListener,
+from .model import (
+    RealtimeModel,
+    RealtimeModelConfig,
+    RealtimeModelListener,
 )
+from .session import RealtimeSession
 
 __all__ = [
     "RealtimeAgent",
@@ -30,11 +40,19 @@
     "RealtimeSession",
     "RealtimeSessionListener",
     "RealtimeSessionListenerFunc",
-    "APIKeyOrKeyFunc",
+    "RealtimeRunConfig",
+    "RealtimeSessionModelSettings",
+    "RealtimeInputAudioTranscriptionConfig",
+    "RealtimeTurnDetectionConfig",
+    "RealtimeAudioFormat",
+    "RealtimeClientMessage",
+    "RealtimeUserInput",
+    "RealtimeUserInputMessage",
+    "RealtimeUserInputText",
     "RealtimeModelName",
-    "RealtimeSessionTransport",
-    "RealtimeTransportListener",
-    "RealtimeTransportConnectionOptions",
+    "RealtimeModel",
+    "RealtimeModelListener",
+    "RealtimeModelConfig",
     "RealtimeSessionEvent",
     "RealtimeAgentStartEvent",
     "RealtimeAgentEndEvent",

src/agents/realtime/config.py

@@ -1,34 +1,44 @@
 from __future__ import annotations
 
-import inspect
 from typing import (
     Any,
-    Callable,
     Literal,
     Union,
 )
 
 from typing_extensions import NotRequired, TypeAlias, TypedDict
 
 from ..model_settings import ToolChoice
-from ..tool import FunctionTool
-from ..util._types import MaybeAwaitable
+from ..tool import Tool
+
+RealtimeModelName: TypeAlias = Union[
+    Literal[
+        "gpt-4o-realtime-preview",
+        "gpt-4o-mini-realtime-preview",
+        "gpt-4o-realtime-preview-2025-06-03",
+        "gpt-4o-realtime-preview-2024-12-17",
+        "gpt-4o-realtime-preview-2024-10-01",
+        "gpt-4o-mini-realtime-preview-2024-12-17",
+    ],
+    str,
+]
+"""The name of a realtime model."""
 
 
 class RealtimeClientMessage(TypedDict):
     type: str  # explicitly required
     other_data: NotRequired[dict[str, Any]]
 
 
-class UserInputText(TypedDict):
+class RealtimeUserInputText(TypedDict):
     type: Literal["input_text"]
     text: str
 
 
 class RealtimeUserInputMessage(TypedDict):
     type: Literal["message"]
     role: Literal["user"]
-    content: list[UserInputText]
+    content: list[RealtimeUserInputText]
 
 
 RealtimeUserInput: TypeAlias = Union[str, RealtimeUserInputMessage]
@@ -55,9 +65,11 @@ class RealtimeTurnDetectionConfig(TypedDict):
     threshold: NotRequired[float]
 
 
-class RealtimeSessionConfig(TypedDict):
-    api_key: NotRequired[APIKeyOrKeyFunc]
-    model: NotRequired[str]
+class RealtimeSessionModelSettings(TypedDict):
+    """Model settings for a realtime model session."""
+
+    model_name: NotRequired[RealtimeModelName]
+
     instructions: NotRequired[str]
     modalities: NotRequired[list[Literal["text", "audio"]]]
     voice: NotRequired[str]
@@ -68,24 +80,13 @@ class RealtimeSessionConfig(TypedDict):
     turn_detection: NotRequired[RealtimeTurnDetectionConfig]
 
     tool_choice: NotRequired[ToolChoice]
-    tools: NotRequired[list[FunctionTool]]
-
-
-APIKeyOrKeyFunc = str | Callable[[], MaybeAwaitable[str]]
-"""Either an API key or a function that returns an API key."""
-
+    tools: NotRequired[list[Tool]]
 
-async def get_api_key(key: APIKeyOrKeyFunc | None) -> str | None:
-    """Get the API key from the key or key function."""
-    if key is None:
-        return None
-    elif isinstance(key, str):
-        return key
 
-    result = key()
-    if inspect.isawaitable(result):
-        return await result
-    return result
+class RealtimeRunConfig(TypedDict):
+    model_settings: NotRequired[RealtimeSessionModelSettings]
 
     # TODO (rm) Add tracing support
     # tracing: NotRequired[RealtimeTracingConfig | None]
+    # TODO (rm) Add guardrail support
+    # TODO (rm) Add history audio storage config

src/agents/realtime/events.py

@@ -7,7 +7,7 @@
 from ..tool import Tool
 from .agent import RealtimeAgent
 from .items import RealtimeItem
-from .transport_events import RealtimeTransportAudioEvent, RealtimeTransportEvent
+from .model_events import RealtimeModelAudioEvent, RealtimeModelEvent
 
 
 @dataclass
@@ -96,7 +96,7 @@ class RealtimeToolEnd:
 class RealtimeRawTransportEvent:
     """Forwards raw events from the transport layer."""
 
-    data: RealtimeTransportEvent
+    data: RealtimeModelEvent
     """The raw data from the transport layer."""
 
     info: RealtimeEventInfo
@@ -119,7 +119,7 @@ class RealtimeAudioEnd:
 class RealtimeAudio:
     """Triggered when the agent generates new audio to be played."""
 
-    audio: RealtimeTransportAudioEvent
+    audio: RealtimeModelAudioEvent
     """The audio event from the transport layer."""
 
     info: RealtimeEventInfo