Merge pull request #3205 from openai/release-please--branches--main--changes--next

release: 2.36.0

Author: apcha-oai

.release-please-manifest.json

@@ -1,3 +1,3 @@
 {
-  ".": "2.35.1"
+  ".": "2.36.0"
 }

.stats.yml

@@ -1,4 +1,4 @@
 configured_endpoints: 233
-openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/openai/openai-84d31411083374ec6cdb4a722f8b8b83c1230741157306b1ca7ba1a3cf246672.yml
-openapi_spec_hash: 051fce676f959b8207e2317225ec4bdc
-config_hash: a2916f18a94ff65c8116ca2fe3256f10
+openapi_spec_url: https://storage.googleapis.com/stainless-sdk-openapi-specs/openai/openai-371f497afe4d6070f6e252e5febbe8f453c7058a8dff0c26a01b4d88442a4ac2.yml
+openapi_spec_hash: d39f46e8fda45f77096448105efd175a
+config_hash: b64135fff1fe9cf4069b9ecf59ae8b07

CHANGELOG.md

@@ -1,5 +1,14 @@
 # Changelog
 
+## 2.36.0 (2026-05-07)
+
+Full Changelog: [v2.35.1...v2.36.0](https://github.com/openai/openai-python/compare/v2.35.1...v2.36.0)
+
+### Features
+
+* **api:** manual updates ([13c639c](https://github.com/openai/openai-python/commit/13c639cc7d57e4fbd4406563511e15eeb88a54b2))
+* **api:** realtime 2 ([8fe0ab8](https://github.com/openai/openai-python/commit/8fe0ab87e67eeb3cc27426b50093845229520f0e))
+
 ## 2.35.1 (2026-05-06)
 
 Full Changelog: [v2.35.0...v2.35.1](https://github.com/openai/openai-python/compare/v2.35.0...v2.35.1)

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "openai"
-version = "2.35.1"
+version = "2.36.0"
 description = "The official Python library for the openai API"
 dynamic = ["readme"]
 license = "Apache-2.0"

src/openai/_version.py

@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "openai"
-__version__ = "2.35.1"  # x-release-please-version
+__version__ = "2.36.0"  # x-release-please-version

src/openai/resources/realtime/api.md

@@ -58,6 +58,8 @@ from openai.types.realtime import (
     RealtimeMcpToolCall,
     RealtimeMcpToolExecutionError,
     RealtimeMcphttpError,
+    RealtimeReasoning,
+    RealtimeReasoningEffort,
     RealtimeResponse,
     RealtimeResponseCreateAudioOutput,
     RealtimeResponseCreateMcpTool,
@@ -77,6 +79,22 @@ from openai.types.realtime import (
     RealtimeTranscriptionSessionAudioInput,
     RealtimeTranscriptionSessionAudioInputTurnDetection,
     RealtimeTranscriptionSessionCreateRequest,
+    RealtimeTranslationClientEvent,
+    RealtimeTranslationClientSecretCreateRequest,
+    RealtimeTranslationClientSecretCreateResponse,
+    RealtimeTranslationInputAudioBufferAppendEvent,
+    RealtimeTranslationInputTranscriptDeltaEvent,
+    RealtimeTranslationOutputAudioDeltaEvent,
+    RealtimeTranslationOutputTranscriptDeltaEvent,
+    RealtimeTranslationServerEvent,
+    RealtimeTranslationSession,
+    RealtimeTranslationSessionCloseEvent,
+    RealtimeTranslationSessionClosedEvent,
+    RealtimeTranslationSessionCreateRequest,
+    RealtimeTranslationSessionCreatedEvent,
+    RealtimeTranslationSessionUpdateEvent,
+    RealtimeTranslationSessionUpdateRequest,
+    RealtimeTranslationSessionUpdatedEvent,
     RealtimeTruncation,
     RealtimeTruncationRetentionRatio,
     ResponseAudioDeltaEvent,
@@ -114,7 +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.
     """