feat(api): realtime 2

Author: stainless-app[bot]

.stats.yml

@@ -1,4 +1,4 @@
 configured_endpoints: 233
-openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/openai/openai-08cb8ed18dfe4a9fa518e278576d3cfe5710cb5c22789cf80826c900569bcf56.yml
-openapi_spec_hash: 20f820c94f54741b75d719f6a7371c12
-config_hash: f291a449469edfe61a28424e548899b2
+openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/openai/openai-371f497afe4d6070f6e252e5febbe8f453c7058a8dff0c26a01b4d88442a4ac2.yml
+openapi_spec_hash: d39f46e8fda45f77096448105efd175a
+config_hash: b64135fff1fe9cf4069b9ecf59ae8b07

src/openai/resources/realtime/api.md

@@ -58,6 +58,8 @@ from openai.types.realtime import (
     RealtimeMcpToolCall,
     RealtimeMcpToolExecutionError,
     RealtimeMcphttpError,
+    RealtimeReasoning,
+    RealtimeReasoningEffort,
     RealtimeResponse,
     RealtimeResponseCreateAudioOutput,
     RealtimeResponseCreateMcpTool,
@@ -130,7 +132,6 @@ Types:
 
 ```python
 from openai.types.realtime import (
-    RealtimeSessionClientSecret,
     RealtimeSessionCreateResponse,
     RealtimeTranscriptionSessionCreateResponse,
     RealtimeTranscriptionSessionTurnDetection,

src/openai/resources/realtime/calls.py

@@ -28,6 +28,7 @@
     call_reject_params,
 )
 from ...types.responses.response_prompt_param import ResponsePromptParam
+from ...types.realtime.realtime_reasoning_param import RealtimeReasoningParam
 from ...types.realtime.realtime_truncation_param import RealtimeTruncationParam
 from ...types.realtime.realtime_audio_config_param import RealtimeAudioConfigParam
 from ...types.realtime.realtime_tools_config_param import RealtimeToolsConfigParam
@@ -121,6 +122,7 @@ def accept(
             Literal[
                 "gpt-realtime",
                 "gpt-realtime-1.5",
+                "gpt-realtime-2",
                 "gpt-realtime-2025-08-28",
                 "gpt-4o-realtime-preview",
                 "gpt-4o-realtime-preview-2024-10-01",
@@ -139,7 +141,9 @@ def accept(
         ]
         | Omit = omit,
         output_modalities: List[Literal["text", "audio"]] | Omit = omit,
+        parallel_tool_calls: bool | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
+        reasoning: RealtimeReasoningParam | Omit = omit,
         tool_choice: RealtimeToolChoiceConfigParam | Omit = omit,
         tools: RealtimeToolsConfigParam | Omit = omit,
         tracing: Optional[RealtimeTracingConfigParam] | Omit = omit,
@@ -188,9 +192,14 @@ def accept(
               can be used to make the model respond with text only. It is not possible to
               request both `text` and `audio` at the same time.
 
+          parallel_tool_calls: Whether the model may call multiple tools in parallel. Only supported by
+              reasoning Realtime models such as `gpt-realtime-2`.
+
           prompt: Reference to a prompt template and its variables.
               [Learn more](https://platform.openai.com/docs/guides/text?api-mode=responses#reusable-prompts).
 
+          reasoning: Configuration for reasoning-capable Realtime models such as `gpt-realtime-2`.
+
           tool_choice: How the model chooses tools. Provide one of the string modes or force a specific
               function/MCP tool.
 
@@ -245,7 +254,9 @@ def accept(
                     "max_output_tokens": max_output_tokens,
                     "model": model,
                     "output_modalities": output_modalities,
+                    "parallel_tool_calls": parallel_tool_calls,
                     "prompt": prompt,
+                    "reasoning": reasoning,
                     "tool_choice": tool_choice,
                     "tools": tools,
                     "tracing": tracing,
@@ -471,6 +482,7 @@ async def accept(
             Literal[
                 "gpt-realtime",
                 "gpt-realtime-1.5",
+                "gpt-realtime-2",
                 "gpt-realtime-2025-08-28",
                 "gpt-4o-realtime-preview",
                 "gpt-4o-realtime-preview-2024-10-01",
@@ -489,7 +501,9 @@ async def accept(
         ]
         | Omit = omit,
         output_modalities: List[Literal["text", "audio"]] | Omit = omit,
+        parallel_tool_calls: bool | Omit = omit,
         prompt: Optional[ResponsePromptParam] | Omit = omit,
+        reasoning: RealtimeReasoningParam | Omit = omit,
         tool_choice: RealtimeToolChoiceConfigParam | Omit = omit,
         tools: RealtimeToolsConfigParam | Omit = omit,
         tracing: Optional[RealtimeTracingConfigParam] | Omit = omit,
@@ -538,9 +552,14 @@ async def accept(
               can be used to make the model respond with text only. It is not possible to
               request both `text` and `audio` at the same time.
 
+          parallel_tool_calls: Whether the model may call multiple tools in parallel. Only supported by
+              reasoning Realtime models such as `gpt-realtime-2`.
+
           prompt: Reference to a prompt template and its variables.
               [Learn more](https://platform.openai.com/docs/guides/text?api-mode=responses#reusable-prompts).
 
+          reasoning: Configuration for reasoning-capable Realtime models such as `gpt-realtime-2`.
+
           tool_choice: How the model chooses tools. Provide one of the string modes or force a specific
               function/MCP tool.
 
@@ -595,7 +614,9 @@ async def accept(
                     "max_output_tokens": max_output_tokens,
                     "model": model,
                     "output_modalities": output_modalities,
+                    "parallel_tool_calls": parallel_tool_calls,
                     "prompt": prompt,
+                    "reasoning": reasoning,
                     "tool_choice": tool_choice,
                     "tools": tools,
                     "tracing": tracing,

src/openai/types/realtime/__init__.py

@@ -9,6 +9,7 @@
 from .call_accept_params import CallAcceptParams as CallAcceptParams
 from .call_create_params import CallCreateParams as CallCreateParams
 from .call_reject_params import CallRejectParams as CallRejectParams
+from .realtime_reasoning import RealtimeReasoning as RealtimeReasoning
 from .audio_transcription import AudioTranscription as AudioTranscription
 from .log_prob_properties import LogProbProperties as LogProbProperties
 from .realtime_truncation import RealtimeTruncation as RealtimeTruncation
@@ -38,11 +39,13 @@
 from .realtime_response_usage import RealtimeResponseUsage as RealtimeResponseUsage
 from .realtime_tracing_config import RealtimeTracingConfig as RealtimeTracingConfig
 from .mcp_list_tools_completed import McpListToolsCompleted as McpListToolsCompleted
+from .realtime_reasoning_param import RealtimeReasoningParam as RealtimeReasoningParam
 from .realtime_response_status import RealtimeResponseStatus as RealtimeResponseStatus
 from .response_mcp_call_failed import ResponseMcpCallFailed as ResponseMcpCallFailed
 from .response_text_done_event import ResponseTextDoneEvent as ResponseTextDoneEvent
 from .audio_transcription_param import AudioTranscriptionParam as AudioTranscriptionParam
 from .rate_limits_updated_event import RateLimitsUpdatedEvent as RateLimitsUpdatedEvent
+from .realtime_reasoning_effort import RealtimeReasoningEffort as RealtimeReasoningEffort
 from .realtime_truncation_param import RealtimeTruncationParam as RealtimeTruncationParam
 from .response_audio_done_event import ResponseAudioDoneEvent as ResponseAudioDoneEvent
 from .response_text_delta_event import ResponseTextDeltaEvent as ResponseTextDeltaEvent
@@ -75,7 +78,6 @@
 from .conversation_item_delete_event import ConversationItemDeleteEvent as ConversationItemDeleteEvent
 from .input_audio_buffer_clear_event import InputAudioBufferClearEvent as InputAudioBufferClearEvent
 from .realtime_mcp_approval_response import RealtimeMcpApprovalResponse as RealtimeMcpApprovalResponse
-from .realtime_session_client_secret import RealtimeSessionClientSecret as RealtimeSessionClientSecret
 from .conversation_item_created_event import ConversationItemCreatedEvent as ConversationItemCreatedEvent
 from .conversation_item_deleted_event import ConversationItemDeletedEvent as ConversationItemDeletedEvent
 from .input_audio_buffer_append_event import InputAudioBufferAppendEvent as InputAudioBufferAppendEvent

src/openai/types/realtime/audio_transcription.py

@@ -9,6 +9,13 @@
 
 
 class AudioTranscription(BaseModel):
+    delay: Optional[Literal["minimal", "low", "medium", "high", "xhigh"]] = None
+    """
+    Controls how long the model waits before emitting transcription text. Higher
+    values can improve transcription accuracy at the cost of latency. Only supported
+    with `gpt-realtime-whisper` in GA Realtime sessions.
+    """
+
     language: Optional[str] = None
     """The language of the input audio.
 
@@ -25,15 +32,16 @@ class AudioTranscription(BaseModel):
             "gpt-4o-mini-transcribe-2025-12-15",
             "gpt-4o-transcribe",
             "gpt-4o-transcribe-diarize",
+            "gpt-realtime-whisper",
         ],
         None,
     ] = None
     """The model to use for transcription.
 
     Current options are `whisper-1`, `gpt-4o-mini-transcribe`,
-    `gpt-4o-mini-transcribe-2025-12-15`, `gpt-4o-transcribe`, and
-    `gpt-4o-transcribe-diarize`. Use `gpt-4o-transcribe-diarize` when you need
-    diarization with speaker labels.
+    `gpt-4o-mini-transcribe-2025-12-15`, `gpt-4o-transcribe`,
+    `gpt-4o-transcribe-diarize`, and `gpt-realtime-whisper`. Use
+    `gpt-4o-transcribe-diarize` when you need diarization with speaker labels.
     """
 
     prompt: Optional[str] = None
@@ -43,4 +51,5 @@ class AudioTranscription(BaseModel):
     [prompt is a list of keywords](https://platform.openai.com/docs/guides/speech-to-text#prompting).
     For `gpt-4o-transcribe` models (excluding `gpt-4o-transcribe-diarize`), the
     prompt is a free text string, for example "expect words related to technology".
+    Prompt is not supported with `gpt-realtime-whisper` in GA Realtime sessions.
     """

src/openai/types/realtime/audio_transcription_param.py

@@ -9,6 +9,13 @@
 
 
 class AudioTranscriptionParam(TypedDict, total=False):
+    delay: Literal["minimal", "low", "medium", "high", "xhigh"]
+    """
+    Controls how long the model waits before emitting transcription text. Higher
+    values can improve transcription accuracy at the cost of latency. Only supported
+    with `gpt-realtime-whisper` in GA Realtime sessions.
+    """
+
     language: str
     """The language of the input audio.
 
@@ -25,14 +32,15 @@ class AudioTranscriptionParam(TypedDict, total=False):
             "gpt-4o-mini-transcribe-2025-12-15",
             "gpt-4o-transcribe",
             "gpt-4o-transcribe-diarize",
+            "gpt-realtime-whisper",
         ],
     ]
     """The model to use for transcription.
 
     Current options are `whisper-1`, `gpt-4o-mini-transcribe`,
-    `gpt-4o-mini-transcribe-2025-12-15`, `gpt-4o-transcribe`, and
-    `gpt-4o-transcribe-diarize`. Use `gpt-4o-transcribe-diarize` when you need
-    diarization with speaker labels.
+    `gpt-4o-mini-transcribe-2025-12-15`, `gpt-4o-transcribe`,
+    `gpt-4o-transcribe-diarize`, and `gpt-realtime-whisper`. Use
+    `gpt-4o-transcribe-diarize` when you need diarization with speaker labels.
     """
 
     prompt: str
@@ -42,4 +50,5 @@ class AudioTranscriptionParam(TypedDict, total=False):
     [prompt is a list of keywords](https://platform.openai.com/docs/guides/speech-to-text#prompting).
     For `gpt-4o-transcribe` models (excluding `gpt-4o-transcribe-diarize`), the
     prompt is a free text string, for example "expect words related to technology".
+    Prompt is not supported with `gpt-realtime-whisper` in GA Realtime sessions.
     """

src/openai/types/realtime/call_accept_params.py

@@ -5,6 +5,7 @@
 from typing import List, Union, Optional
 from typing_extensions import Literal, Required, TypedDict
 
+from .realtime_reasoning_param import RealtimeReasoningParam
 from .realtime_truncation_param import RealtimeTruncationParam
 from .realtime_audio_config_param import RealtimeAudioConfigParam
 from .realtime_tools_config_param import RealtimeToolsConfigParam
@@ -57,6 +58,7 @@ class CallAcceptParams(TypedDict, total=False):
         Literal[
             "gpt-realtime",
             "gpt-realtime-1.5",
+            "gpt-realtime-2",
             "gpt-realtime-2025-08-28",
             "gpt-4o-realtime-preview",
             "gpt-4o-realtime-preview-2024-10-01",
@@ -83,12 +85,21 @@ class CallAcceptParams(TypedDict, total=False):
     only. It is not possible to request both `text` and `audio` at the same time.
     """
 
+    parallel_tool_calls: bool
+    """Whether the model may call multiple tools in parallel.
+
+    Only supported by reasoning Realtime models such as `gpt-realtime-2`.
+    """
+
     prompt: Optional[ResponsePromptParam]
     """
     Reference to a prompt template and its variables.
     [Learn more](https://platform.openai.com/docs/guides/text?api-mode=responses#reusable-prompts).
     """
 
+    reasoning: RealtimeReasoningParam
+    """Configuration for reasoning-capable Realtime models such as `gpt-realtime-2`."""
+
     tool_choice: RealtimeToolChoiceConfigParam
     """How the model chooses tools.
 

src/openai/types/realtime/realtime_audio_config_input.py

@@ -67,4 +67,7 @@ class RealtimeAudioConfigInput(BaseModel):
     trails off with "uhhm", the model will score a low probability of turn end and
     wait longer for the user to continue speaking. This can be useful for more
     natural conversations, but may have a higher latency.
+
+    For `gpt-realtime-whisper` transcription sessions, turn detection must be set to
+    `null`; VAD is not supported.
     """

src/openai/types/realtime/realtime_audio_config_input_param.py

@@ -69,4 +69,7 @@ class RealtimeAudioConfigInputParam(TypedDict, total=False):
     trails off with "uhhm", the model will score a low probability of turn end and
     wait longer for the user to continue speaking. This can be useful for more
     natural conversations, but may have a higher latency.
+
+    For `gpt-realtime-whisper` transcription sessions, turn detection must be set to
+    `null`; VAD is not supported.
     """

src/openai/types/realtime/realtime_reasoning.py

@@ -0,0 +1,18 @@
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Optional
+
+from ..._models import BaseModel
+from .realtime_reasoning_effort import RealtimeReasoningEffort
+
+__all__ = ["RealtimeReasoning"]
+
+
+class RealtimeReasoning(BaseModel):
+    """Configuration for reasoning-capable Realtime models such as `gpt-realtime-2`."""
+
+    effort: Optional[RealtimeReasoningEffort] = None
+    """
+    Constrains effort on reasoning for reasoning-capable Realtime models such as
+    `gpt-realtime-2`.
+    """