openai 0.36.1 → 0.37.0

data/lib/openai/models/evals/run_retrieve_response.rb

@@ -316,9 +316,9 @@ module OpenAI
                 # @!attribute reasoning_effort
                 #   Constrains effort on reasoning for
                 #   [reasoning models](https://platform.openai.com/docs/guides/reasoning). Currently
-                #   supported values are `none`, `minimal`, `low`, `medium`, and `high`. Reducing
-                #   reasoning effort can result in faster responses and fewer tokens used on
-                #   reasoning in a response.
+                #   supported values are `none`, `minimal`, `low`, `medium`, `high`, and `xhigh`.
+                #   Reducing reasoning effort can result in faster responses and fewer tokens used
+                #   on reasoning in a response.
                 #
                 #   - `gpt-5.1` defaults to `none`, which does not perform reasoning. The supported
                 #     reasoning values for `gpt-5.1` are `none`, `low`, `medium`, and `high`. Tool
@@ -326,6 +326,7 @@ module OpenAI
                 #   - All models before `gpt-5.1` default to `medium` reasoning effort, and do not
                 #     support `none`.
                 #   - The `gpt-5-pro` model defaults to (and only supports) `high` reasoning effort.
+                #   - `xhigh` is currently only supported for `gpt-5.1-codex-max`.
                 #
                 #   @return [Symbol, OpenAI::Models::ReasoningEffort, nil]
                 optional :reasoning_effort, enum: -> { OpenAI::ReasoningEffort }, nil?: true
@@ -668,9 +669,9 @@ module OpenAI
               # @!attribute reasoning_effort
               #   Constrains effort on reasoning for
               #   [reasoning models](https://platform.openai.com/docs/guides/reasoning). Currently
-              #   supported values are `none`, `minimal`, `low`, `medium`, and `high`. Reducing
-              #   reasoning effort can result in faster responses and fewer tokens used on
-              #   reasoning in a response.
+              #   supported values are `none`, `minimal`, `low`, `medium`, `high`, and `xhigh`.
+              #   Reducing reasoning effort can result in faster responses and fewer tokens used
+              #   on reasoning in a response.
               #
               #   - `gpt-5.1` defaults to `none`, which does not perform reasoning. The supported
               #     reasoning values for `gpt-5.1` are `none`, `low`, `medium`, and `high`. Tool
@@ -678,6 +679,7 @@ module OpenAI
               #   - All models before `gpt-5.1` default to `medium` reasoning effort, and do not
               #     support `none`.
               #   - The `gpt-5-pro` model defaults to (and only supports) `high` reasoning effort.
+              #   - `xhigh` is currently only supported for `gpt-5.1-codex-max`.
               #
               #   @return [Symbol, OpenAI::Models::ReasoningEffort, nil]
               optional :reasoning_effort, enum: -> { OpenAI::ReasoningEffort }, nil?: true

data/lib/openai/models/graders/score_model_grader.rb

@@ -222,9 +222,9 @@ module OpenAI
           # @!attribute reasoning_effort
           #   Constrains effort on reasoning for
           #   [reasoning models](https://platform.openai.com/docs/guides/reasoning). Currently
-          #   supported values are `none`, `minimal`, `low`, `medium`, and `high`. Reducing
-          #   reasoning effort can result in faster responses and fewer tokens used on
-          #   reasoning in a response.
+          #   supported values are `none`, `minimal`, `low`, `medium`, `high`, and `xhigh`.
+          #   Reducing reasoning effort can result in faster responses and fewer tokens used
+          #   on reasoning in a response.
           #
           #   - `gpt-5.1` defaults to `none`, which does not perform reasoning. The supported
           #     reasoning values for `gpt-5.1` are `none`, `low`, `medium`, and `high`. Tool
@@ -232,6 +232,7 @@ module OpenAI
           #   - All models before `gpt-5.1` default to `medium` reasoning effort, and do not
           #     support `none`.
           #   - The `gpt-5-pro` model defaults to (and only supports) `high` reasoning effort.
+          #   - `xhigh` is currently only supported for `gpt-5.1-codex-max`.
           #
           #   @return [Symbol, OpenAI::Models::ReasoningEffort, nil]
           optional :reasoning_effort, enum: -> { OpenAI::ReasoningEffort }, nil?: true

data/lib/openai/models/realtime/input_audio_buffer_dtmf_event_received_event.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+module OpenAI
+  module Models
+    module Realtime
+      class InputAudioBufferDtmfEventReceivedEvent < OpenAI::Internal::Type::BaseModel
+        # @!attribute event
+        #   The telephone keypad that was pressed by the user.
+        #
+        #   @return [String]
+        required :event, String
+
+        # @!attribute received_at
+        #   UTC Unix Timestamp when DTMF Event was received by server.
+        #
+        #   @return [Integer]
+        required :received_at, Integer
+
+        # @!attribute type
+        #   The event type, must be `input_audio_buffer.dtmf_event_received`.
+        #
+        #   @return [Symbol, :"input_audio_buffer.dtmf_event_received"]
+        required :type, const: :"input_audio_buffer.dtmf_event_received"
+
+        # @!method initialize(event:, received_at:, type: :"input_audio_buffer.dtmf_event_received")
+        #   Some parameter documentations has been truncated, see
+        #   {OpenAI::Models::Realtime::InputAudioBufferDtmfEventReceivedEvent} for more
+        #   details.
+        #
+        #   **SIP Only:** Returned when an DTMF event is received. A DTMF event is a message
+        #   that represents a telephone keypad press (0–9, \*, #, A–D). The `event` property
+        #   is the keypad that the user press. The `received_at` is the UTC Unix Timestamp
+        #   that the server received the event.
+        #
+        #   @param event [String] The telephone keypad that was pressed by the user.
+        #
+        #   @param received_at [Integer] UTC Unix Timestamp when DTMF Event was received by server.
+        #
+        #   @param type [Symbol, :"input_audio_buffer.dtmf_event_received"] The event type, must be `input_audio_buffer.dtmf_event_received`.
+      end
+    end
+  end
+end

data/lib/openai/models/realtime/output_audio_buffer_clear_event.rb

@@ -17,10 +17,10 @@ module OpenAI
         optional :event_id, String
 
         # @!method initialize(event_id: nil, type: :"output_audio_buffer.clear")
-        #   **WebRTC Only:** Emit to cut off the current audio response. This will trigger
-        #   the server to stop generating audio and emit a `output_audio_buffer.cleared`
-        #   event. This event should be preceded by a `response.cancel` client event to stop
-        #   the generation of the current response.
+        #   **WebRTC/SIP Only:** Emit to cut off the current audio response. This will
+        #   trigger the server to stop generating audio and emit a
+        #   `output_audio_buffer.cleared` event. This event should be preceded by a
+        #   `response.cancel` client event to stop the generation of the current response.
         #   [Learn more](https://platform.openai.com/docs/guides/realtime-conversations#client-and-server-events-for-audio-in-webrtc).
         #
         #   @param event_id [String] The unique ID of the client event used for error handling.

data/lib/openai/models/realtime/realtime_audio_input_turn_detection.rb

@@ -36,7 +36,11 @@ module OpenAI
 
           # @!attribute create_response
           #   Whether or not to automatically generate a response when a VAD stop event
-          #   occurs.
+          #   occurs. If `interrupt_response` is set to `false` this may fail to create a
+          #   response if the model is already responding.
+          #
+          #   If both `create_response` and `interrupt_response` are set to `false`, the model
+          #   will never respond automatically but VAD events will still be emitted.
           #
           #   @return [Boolean, nil]
           optional :create_response, OpenAI::Internal::Type::Boolean
@@ -59,9 +63,13 @@ module OpenAI
           optional :idle_timeout_ms, Integer, nil?: true
 
           # @!attribute interrupt_response
-          #   Whether or not to automatically interrupt any ongoing response with output to
-          #   the default conversation (i.e. `conversation` of `auto`) when a VAD start event
-          #   occurs.
+          #   Whether or not to automatically interrupt (cancel) any ongoing response with
+          #   output to the default conversation (i.e. `conversation` of `auto`) when a VAD
+          #   start event occurs. If `true` then the response will be cancelled, otherwise it
+          #   will continue until complete.
+          #
+          #   If both `create_response` and `interrupt_response` are set to `false`, the model
+          #   will never respond automatically but VAD events will still be emitted.
           #
           #   @return [Boolean, nil]
           optional :interrupt_response, OpenAI::Internal::Type::Boolean
@@ -101,7 +109,7 @@ module OpenAI
           #
           #   @param idle_timeout_ms [Integer, nil] Optional timeout after which a model response will be triggered automatically. T
           #
-          #   @param interrupt_response [Boolean] Whether or not to automatically interrupt any ongoing response with output to th
+          #   @param interrupt_response [Boolean] Whether or not to automatically interrupt (cancel) any ongoing response with out
           #
           #   @param prefix_padding_ms [Integer] Used only for `server_vad` mode. Amount of audio to include before the VAD detec
           #

data/lib/openai/models/realtime/realtime_client_event.rb

@@ -62,7 +62,7 @@ module OpenAI
         # respond with an `input_audio_buffer.cleared` event.
         variant :"input_audio_buffer.clear", -> { OpenAI::Realtime::InputAudioBufferClearEvent }
 
-        # **WebRTC Only:** Emit to cut off the current audio response. This will trigger the server to
+        # **WebRTC/SIP Only:** Emit to cut off the current audio response. This will trigger the server to
         # stop generating audio and emit a `output_audio_buffer.cleared` event. This
         # event should be preceded by a `response.cancel` client event to stop the
         # generation of the current response.

data/lib/openai/models/realtime/realtime_server_event.rb

@@ -78,6 +78,13 @@ module OpenAI
         # will also be sent to the client.
         variant :"input_audio_buffer.committed", -> { OpenAI::Realtime::InputAudioBufferCommittedEvent }
 
+        # **SIP Only:** Returned when an DTMF event is received. A DTMF event is a message that
+        # represents a telephone keypad press (0–9, *, #, A–D). The `event` property
+        # is the keypad that the user press. The `received_at` is the UTC Unix Timestamp
+        # that the server received the event.
+        variant :"input_audio_buffer.dtmf_event_received",
+                -> { OpenAI::Realtime::InputAudioBufferDtmfEventReceivedEvent }
+
         # Sent by the server when in `server_vad` mode to indicate that speech has been
         # detected in the audio buffer. This can happen any time audio is added to the
         # buffer (unless speech is already detected). The client may want to use this
@@ -173,21 +180,21 @@ module OpenAI
         # there is an error.
         variant :"session.updated", -> { OpenAI::Realtime::SessionUpdatedEvent }
 
-        # **WebRTC Only:** Emitted when the server begins streaming audio to the client. This event is
+        # **WebRTC/SIP Only:** Emitted when the server begins streaming audio to the client. This event is
         # emitted after an audio content part has been added (`response.content_part.added`)
         # to the response.
         # [Learn more](https://platform.openai.com/docs/guides/realtime-conversations#client-and-server-events-for-audio-in-webrtc).
         variant :"output_audio_buffer.started",
                 -> { OpenAI::Realtime::RealtimeServerEvent::OutputAudioBufferStarted }
 
-        # **WebRTC Only:** Emitted when the output audio buffer has been completely drained on the server,
+        # **WebRTC/SIP Only:** Emitted when the output audio buffer has been completely drained on the server,
         # and no more audio is forthcoming. This event is emitted after the full response
         # data has been sent to the client (`response.done`).
         # [Learn more](https://platform.openai.com/docs/guides/realtime-conversations#client-and-server-events-for-audio-in-webrtc).
         variant :"output_audio_buffer.stopped",
                 -> { OpenAI::Realtime::RealtimeServerEvent::OutputAudioBufferStopped }
 
-        # **WebRTC Only:** Emitted when the output audio buffer is cleared. This happens either in VAD
+        # **WebRTC/SIP Only:** Emitted when the output audio buffer is cleared. This happens either in VAD
         # mode when the user has interrupted (`input_audio_buffer.speech_started`),
         # or when the client has emitted the `output_audio_buffer.clear` event to manually
         # cut off the current audio response.
@@ -304,8 +311,8 @@ module OpenAI
           required :type, const: :"output_audio_buffer.started"
 
           # @!method initialize(event_id:, response_id:, type: :"output_audio_buffer.started")
-          #   **WebRTC Only:** Emitted when the server begins streaming audio to the client.
-          #   This event is emitted after an audio content part has been added
+          #   **WebRTC/SIP Only:** Emitted when the server begins streaming audio to the
+          #   client. This event is emitted after an audio content part has been added
           #   (`response.content_part.added`) to the response.
           #   [Learn more](https://platform.openai.com/docs/guides/realtime-conversations#client-and-server-events-for-audio-in-webrtc).
           #
@@ -336,7 +343,7 @@ module OpenAI
           required :type, const: :"output_audio_buffer.stopped"
 
           # @!method initialize(event_id:, response_id:, type: :"output_audio_buffer.stopped")
-          #   **WebRTC Only:** Emitted when the output audio buffer has been completely
+          #   **WebRTC/SIP Only:** Emitted when the output audio buffer has been completely
           #   drained on the server, and no more audio is forthcoming. This event is emitted
           #   after the full response data has been sent to the client (`response.done`).
           #   [Learn more](https://platform.openai.com/docs/guides/realtime-conversations#client-and-server-events-for-audio-in-webrtc).
@@ -368,8 +375,8 @@ module OpenAI
           required :type, const: :"output_audio_buffer.cleared"
 
           # @!method initialize(event_id:, response_id:, type: :"output_audio_buffer.cleared")
-          #   **WebRTC Only:** Emitted when the output audio buffer is cleared. This happens
-          #   either in VAD mode when the user has interrupted
+          #   **WebRTC/SIP Only:** Emitted when the output audio buffer is cleared. This
+          #   happens either in VAD mode when the user has interrupted
           #   (`input_audio_buffer.speech_started`), or when the client has emitted the
           #   `output_audio_buffer.clear` event to manually cut off the current audio
           #   response.
@@ -383,7 +390,7 @@ module OpenAI
         end
 
         # @!method self.variants
-        #   @return [Array(OpenAI::Models::Realtime::ConversationCreatedEvent, OpenAI::Models::Realtime::ConversationItemCreatedEvent, OpenAI::Models::Realtime::ConversationItemDeletedEvent, OpenAI::Models::Realtime::ConversationItemInputAudioTranscriptionCompletedEvent, OpenAI::Models::Realtime::ConversationItemInputAudioTranscriptionDeltaEvent, OpenAI::Models::Realtime::ConversationItemInputAudioTranscriptionFailedEvent, OpenAI::Models::Realtime::RealtimeServerEvent::ConversationItemRetrieved, OpenAI::Models::Realtime::ConversationItemTruncatedEvent, OpenAI::Models::Realtime::RealtimeErrorEvent, OpenAI::Models::Realtime::InputAudioBufferClearedEvent, OpenAI::Models::Realtime::InputAudioBufferCommittedEvent, OpenAI::Models::Realtime::InputAudioBufferSpeechStartedEvent, OpenAI::Models::Realtime::InputAudioBufferSpeechStoppedEvent, OpenAI::Models::Realtime::RateLimitsUpdatedEvent, OpenAI::Models::Realtime::ResponseAudioDeltaEvent, OpenAI::Models::Realtime::ResponseAudioDoneEvent, OpenAI::Models::Realtime::ResponseAudioTranscriptDeltaEvent, OpenAI::Models::Realtime::ResponseAudioTranscriptDoneEvent, OpenAI::Models::Realtime::ResponseContentPartAddedEvent, OpenAI::Models::Realtime::ResponseContentPartDoneEvent, OpenAI::Models::Realtime::ResponseCreatedEvent, OpenAI::Models::Realtime::ResponseDoneEvent, OpenAI::Models::Realtime::ResponseFunctionCallArgumentsDeltaEvent, OpenAI::Models::Realtime::ResponseFunctionCallArgumentsDoneEvent, OpenAI::Models::Realtime::ResponseOutputItemAddedEvent, OpenAI::Models::Realtime::ResponseOutputItemDoneEvent, OpenAI::Models::Realtime::ResponseTextDeltaEvent, OpenAI::Models::Realtime::ResponseTextDoneEvent, OpenAI::Models::Realtime::SessionCreatedEvent, OpenAI::Models::Realtime::SessionUpdatedEvent, OpenAI::Models::Realtime::RealtimeServerEvent::OutputAudioBufferStarted, OpenAI::Models::Realtime::RealtimeServerEvent::OutputAudioBufferStopped, OpenAI::Models::Realtime::RealtimeServerEvent::OutputAudioBufferCleared, OpenAI::Models::Realtime::ConversationItemAdded, OpenAI::Models::Realtime::ConversationItemDone, OpenAI::Models::Realtime::InputAudioBufferTimeoutTriggered, OpenAI::Models::Realtime::ConversationItemInputAudioTranscriptionSegment, OpenAI::Models::Realtime::McpListToolsInProgress, OpenAI::Models::Realtime::McpListToolsCompleted, OpenAI::Models::Realtime::McpListToolsFailed, OpenAI::Models::Realtime::ResponseMcpCallArgumentsDelta, OpenAI::Models::Realtime::ResponseMcpCallArgumentsDone, OpenAI::Models::Realtime::ResponseMcpCallInProgress, OpenAI::Models::Realtime::ResponseMcpCallCompleted, OpenAI::Models::Realtime::ResponseMcpCallFailed)]
+        #   @return [Array(OpenAI::Models::Realtime::ConversationCreatedEvent, OpenAI::Models::Realtime::ConversationItemCreatedEvent, OpenAI::Models::Realtime::ConversationItemDeletedEvent, OpenAI::Models::Realtime::ConversationItemInputAudioTranscriptionCompletedEvent, OpenAI::Models::Realtime::ConversationItemInputAudioTranscriptionDeltaEvent, OpenAI::Models::Realtime::ConversationItemInputAudioTranscriptionFailedEvent, OpenAI::Models::Realtime::RealtimeServerEvent::ConversationItemRetrieved, OpenAI::Models::Realtime::ConversationItemTruncatedEvent, OpenAI::Models::Realtime::RealtimeErrorEvent, OpenAI::Models::Realtime::InputAudioBufferClearedEvent, OpenAI::Models::Realtime::InputAudioBufferCommittedEvent, OpenAI::Models::Realtime::InputAudioBufferDtmfEventReceivedEvent, OpenAI::Models::Realtime::InputAudioBufferSpeechStartedEvent, OpenAI::Models::Realtime::InputAudioBufferSpeechStoppedEvent, OpenAI::Models::Realtime::RateLimitsUpdatedEvent, OpenAI::Models::Realtime::ResponseAudioDeltaEvent, OpenAI::Models::Realtime::ResponseAudioDoneEvent, OpenAI::Models::Realtime::ResponseAudioTranscriptDeltaEvent, OpenAI::Models::Realtime::ResponseAudioTranscriptDoneEvent, OpenAI::Models::Realtime::ResponseContentPartAddedEvent, OpenAI::Models::Realtime::ResponseContentPartDoneEvent, OpenAI::Models::Realtime::ResponseCreatedEvent, OpenAI::Models::Realtime::ResponseDoneEvent, OpenAI::Models::Realtime::ResponseFunctionCallArgumentsDeltaEvent, OpenAI::Models::Realtime::ResponseFunctionCallArgumentsDoneEvent, OpenAI::Models::Realtime::ResponseOutputItemAddedEvent, OpenAI::Models::Realtime::ResponseOutputItemDoneEvent, OpenAI::Models::Realtime::ResponseTextDeltaEvent, OpenAI::Models::Realtime::ResponseTextDoneEvent, OpenAI::Models::Realtime::SessionCreatedEvent, OpenAI::Models::Realtime::SessionUpdatedEvent, OpenAI::Models::Realtime::RealtimeServerEvent::OutputAudioBufferStarted, OpenAI::Models::Realtime::RealtimeServerEvent::OutputAudioBufferStopped, OpenAI::Models::Realtime::RealtimeServerEvent::OutputAudioBufferCleared, OpenAI::Models::Realtime::ConversationItemAdded, OpenAI::Models::Realtime::ConversationItemDone, OpenAI::Models::Realtime::InputAudioBufferTimeoutTriggered, OpenAI::Models::Realtime::ConversationItemInputAudioTranscriptionSegment, OpenAI::Models::Realtime::McpListToolsInProgress, OpenAI::Models::Realtime::McpListToolsCompleted, OpenAI::Models::Realtime::McpListToolsFailed, OpenAI::Models::Realtime::ResponseMcpCallArgumentsDelta, OpenAI::Models::Realtime::ResponseMcpCallArgumentsDone, OpenAI::Models::Realtime::ResponseMcpCallInProgress, OpenAI::Models::Realtime::ResponseMcpCallCompleted, OpenAI::Models::Realtime::ResponseMcpCallFailed)]
       end
     end
   end

data/lib/openai/models/realtime/realtime_session.rb

@@ -443,7 +443,11 @@ module OpenAI
 
             # @!attribute create_response
             #   Whether or not to automatically generate a response when a VAD stop event
-            #   occurs.
+            #   occurs. If `interrupt_response` is set to `false` this may fail to create a
+            #   response if the model is already responding.
+            #
+            #   If both `create_response` and `interrupt_response` are set to `false`, the model
+            #   will never respond automatically but VAD events will still be emitted.
             #
             #   @return [Boolean, nil]
             optional :create_response, OpenAI::Internal::Type::Boolean
@@ -466,9 +470,13 @@ module OpenAI
             optional :idle_timeout_ms, Integer, nil?: true
 
             # @!attribute interrupt_response
-            #   Whether or not to automatically interrupt any ongoing response with output to
-            #   the default conversation (i.e. `conversation` of `auto`) when a VAD start event
-            #   occurs.
+            #   Whether or not to automatically interrupt (cancel) any ongoing response with
+            #   output to the default conversation (i.e. `conversation` of `auto`) when a VAD
+            #   start event occurs. If `true` then the response will be cancelled, otherwise it
+            #   will continue until complete.
+            #
+            #   If both `create_response` and `interrupt_response` are set to `false`, the model
+            #   will never respond automatically but VAD events will still be emitted.
             #
             #   @return [Boolean, nil]
             optional :interrupt_response, OpenAI::Internal::Type::Boolean
@@ -508,7 +516,7 @@ module OpenAI
             #
             #   @param idle_timeout_ms [Integer, nil] Optional timeout after which a model response will be triggered automatically. T
             #
-            #   @param interrupt_response [Boolean] Whether or not to automatically interrupt any ongoing response with output to th
+            #   @param interrupt_response [Boolean] Whether or not to automatically interrupt (cancel) any ongoing response with out
             #
             #   @param prefix_padding_ms [Integer] Used only for `server_vad` mode. Amount of audio to include before the VAD detec
             #

data/lib/openai/models/realtime/realtime_session_create_request.rb

@@ -102,15 +102,20 @@ module OpenAI
         #   limit, the conversation be truncated, meaning messages (starting from the
         #   oldest) will not be included in the model's context. A 32k context model with
         #   4,096 max output tokens can only include 28,224 tokens in the context before
-        #   truncation occurs. Clients can configure truncation behavior to truncate with a
-        #   lower max token limit, which is an effective way to control token usage and
-        #   cost. Truncation will reduce the number of cached tokens on the next turn
-        #   (busting the cache), since messages are dropped from the beginning of the
-        #   context. However, clients can also configure truncation to retain messages up to
-        #   a fraction of the maximum context size, which will reduce the need for future
-        #   truncations and thus improve the cache rate. Truncation can be disabled
-        #   entirely, which means the server will never truncate but would instead return an
-        #   error if the conversation exceeds the model's input token limit.
+        #   truncation occurs.
+        #
+        #   Clients can configure truncation behavior to truncate with a lower max token
+        #   limit, which is an effective way to control token usage and cost.
+        #
+        #   Truncation will reduce the number of cached tokens on the next turn (busting the
+        #   cache), since messages are dropped from the beginning of the context. However,
+        #   clients can also configure truncation to retain messages up to a fraction of the
+        #   maximum context size, which will reduce the need for future truncations and thus
+        #   improve the cache rate.
+        #
+        #   Truncation can be disabled entirely, which means the server will never truncate
+        #   but would instead return an error if the conversation exceeds the model's input
+        #   token limit.
         #
         #   @return [Symbol, OpenAI::Models::Realtime::RealtimeTruncation::RealtimeTruncationStrategy, OpenAI::Models::Realtime::RealtimeTruncationRetentionRatio, nil]
         optional :truncation, union: -> { OpenAI::Realtime::RealtimeTruncation }

data/lib/openai/models/realtime/realtime_session_create_response.rb

@@ -110,15 +110,20 @@ module OpenAI
         #   limit, the conversation be truncated, meaning messages (starting from the
         #   oldest) will not be included in the model's context. A 32k context model with
         #   4,096 max output tokens can only include 28,224 tokens in the context before
-        #   truncation occurs. Clients can configure truncation behavior to truncate with a
-        #   lower max token limit, which is an effective way to control token usage and
-        #   cost. Truncation will reduce the number of cached tokens on the next turn
-        #   (busting the cache), since messages are dropped from the beginning of the
-        #   context. However, clients can also configure truncation to retain messages up to
-        #   a fraction of the maximum context size, which will reduce the need for future
-        #   truncations and thus improve the cache rate. Truncation can be disabled
-        #   entirely, which means the server will never truncate but would instead return an
-        #   error if the conversation exceeds the model's input token limit.
+        #   truncation occurs.
+        #
+        #   Clients can configure truncation behavior to truncate with a lower max token
+        #   limit, which is an effective way to control token usage and cost.
+        #
+        #   Truncation will reduce the number of cached tokens on the next turn (busting the
+        #   cache), since messages are dropped from the beginning of the context. However,
+        #   clients can also configure truncation to retain messages up to a fraction of the
+        #   maximum context size, which will reduce the need for future truncations and thus
+        #   improve the cache rate.
+        #
+        #   Truncation can be disabled entirely, which means the server will never truncate
+        #   but would instead return an error if the conversation exceeds the model's input
+        #   token limit.
         #
         #   @return [Symbol, OpenAI::Models::Realtime::RealtimeTruncation::RealtimeTruncationStrategy, OpenAI::Models::Realtime::RealtimeTruncationRetentionRatio, nil]
         optional :truncation, union: -> { OpenAI::Realtime::RealtimeTruncation }
@@ -302,7 +307,11 @@ module OpenAI
 
                 # @!attribute create_response
                 #   Whether or not to automatically generate a response when a VAD stop event
-                #   occurs.
+                #   occurs. If `interrupt_response` is set to `false` this may fail to create a
+                #   response if the model is already responding.
+                #
+                #   If both `create_response` and `interrupt_response` are set to `false`, the model
+                #   will never respond automatically but VAD events will still be emitted.
                 #
                 #   @return [Boolean, nil]
                 optional :create_response, OpenAI::Internal::Type::Boolean
@@ -325,9 +334,13 @@ module OpenAI
                 optional :idle_timeout_ms, Integer, nil?: true
 
                 # @!attribute interrupt_response
-                #   Whether or not to automatically interrupt any ongoing response with output to
-                #   the default conversation (i.e. `conversation` of `auto`) when a VAD start event
-                #   occurs.
+                #   Whether or not to automatically interrupt (cancel) any ongoing response with
+                #   output to the default conversation (i.e. `conversation` of `auto`) when a VAD
+                #   start event occurs. If `true` then the response will be cancelled, otherwise it
+                #   will continue until complete.
+                #
+                #   If both `create_response` and `interrupt_response` are set to `false`, the model
+                #   will never respond automatically but VAD events will still be emitted.
                 #
                 #   @return [Boolean, nil]
                 optional :interrupt_response, OpenAI::Internal::Type::Boolean
@@ -367,7 +380,7 @@ module OpenAI
                 #
                 #   @param idle_timeout_ms [Integer, nil] Optional timeout after which a model response will be triggered automatically. T
                 #
-                #   @param interrupt_response [Boolean] Whether or not to automatically interrupt any ongoing response with output to th
+                #   @param interrupt_response [Boolean] Whether or not to automatically interrupt (cancel) any ongoing response with out
                 #
                 #   @param prefix_padding_ms [Integer] Used only for `server_vad` mode. Amount of audio to include before the VAD detec
                 #

data/lib/openai/models/realtime/realtime_transcription_session_audio_input_turn_detection.rb

@@ -38,7 +38,11 @@ module OpenAI
 
           # @!attribute create_response
           #   Whether or not to automatically generate a response when a VAD stop event
-          #   occurs.
+          #   occurs. If `interrupt_response` is set to `false` this may fail to create a
+          #   response if the model is already responding.
+          #
+          #   If both `create_response` and `interrupt_response` are set to `false`, the model
+          #   will never respond automatically but VAD events will still be emitted.
           #
           #   @return [Boolean, nil]
           optional :create_response, OpenAI::Internal::Type::Boolean
@@ -61,9 +65,13 @@ module OpenAI
           optional :idle_timeout_ms, Integer, nil?: true
 
           # @!attribute interrupt_response
-          #   Whether or not to automatically interrupt any ongoing response with output to
-          #   the default conversation (i.e. `conversation` of `auto`) when a VAD start event
-          #   occurs.
+          #   Whether or not to automatically interrupt (cancel) any ongoing response with
+          #   output to the default conversation (i.e. `conversation` of `auto`) when a VAD
+          #   start event occurs. If `true` then the response will be cancelled, otherwise it
+          #   will continue until complete.
+          #
+          #   If both `create_response` and `interrupt_response` are set to `false`, the model
+          #   will never respond automatically but VAD events will still be emitted.
           #
           #   @return [Boolean, nil]
           optional :interrupt_response, OpenAI::Internal::Type::Boolean
@@ -103,7 +111,7 @@ module OpenAI
           #
           #   @param idle_timeout_ms [Integer, nil] Optional timeout after which a model response will be triggered automatically. T
           #
-          #   @param interrupt_response [Boolean] Whether or not to automatically interrupt any ongoing response with output to th
+          #   @param interrupt_response [Boolean] Whether or not to automatically interrupt (cancel) any ongoing response with out
           #
           #   @param prefix_padding_ms [Integer] Used only for `server_vad` mode. Amount of audio to include before the VAD detec
           #

data/lib/openai/models/realtime/realtime_truncation.rb

@@ -7,15 +7,20 @@ module OpenAI
       # limit, the conversation be truncated, meaning messages (starting from the
       # oldest) will not be included in the model's context. A 32k context model with
       # 4,096 max output tokens can only include 28,224 tokens in the context before
-      # truncation occurs. Clients can configure truncation behavior to truncate with a
-      # lower max token limit, which is an effective way to control token usage and
-      # cost. Truncation will reduce the number of cached tokens on the next turn
-      # (busting the cache), since messages are dropped from the beginning of the
-      # context. However, clients can also configure truncation to retain messages up to
-      # a fraction of the maximum context size, which will reduce the need for future
-      # truncations and thus improve the cache rate. Truncation can be disabled
-      # entirely, which means the server will never truncate but would instead return an
-      # error if the conversation exceeds the model's input token limit.
+      # truncation occurs.
+      #
+      # Clients can configure truncation behavior to truncate with a lower max token
+      # limit, which is an effective way to control token usage and cost.
+      #
+      # Truncation will reduce the number of cached tokens on the next turn (busting the
+      # cache), since messages are dropped from the beginning of the context. However,
+      # clients can also configure truncation to retain messages up to a fraction of the
+      # maximum context size, which will reduce the need for future truncations and thus
+      # improve the cache rate.
+      #
+      # Truncation can be disabled entirely, which means the server will never truncate
+      # but would instead return an error if the conversation exceeds the model's input
+      # token limit.
       module RealtimeTruncation
         extend OpenAI::Internal::Type::Union
 

data/lib/openai/models/reasoning.rb

@@ -6,9 +6,9 @@ module OpenAI
       # @!attribute effort
       #   Constrains effort on reasoning for
       #   [reasoning models](https://platform.openai.com/docs/guides/reasoning). Currently
-      #   supported values are `none`, `minimal`, `low`, `medium`, and `high`. Reducing
-      #   reasoning effort can result in faster responses and fewer tokens used on
-      #   reasoning in a response.
+      #   supported values are `none`, `minimal`, `low`, `medium`, `high`, and `xhigh`.
+      #   Reducing reasoning effort can result in faster responses and fewer tokens used
+      #   on reasoning in a response.
       #
       #   - `gpt-5.1` defaults to `none`, which does not perform reasoning. The supported
       #     reasoning values for `gpt-5.1` are `none`, `low`, `medium`, and `high`. Tool
@@ -16,6 +16,7 @@ module OpenAI
       #   - All models before `gpt-5.1` default to `medium` reasoning effort, and do not
       #     support `none`.
       #   - The `gpt-5-pro` model defaults to (and only supports) `high` reasoning effort.
+      #   - `xhigh` is currently only supported for `gpt-5.1-codex-max`.
       #
       #   @return [Symbol, OpenAI::Models::ReasoningEffort, nil]
       optional :effort, enum: -> { OpenAI::ReasoningEffort }, nil?: true

data/lib/openai/models/reasoning_effort.rb

@@ -4,9 +4,9 @@ module OpenAI
   module Models
     # Constrains effort on reasoning for
     # [reasoning models](https://platform.openai.com/docs/guides/reasoning). Currently
-    # supported values are `none`, `minimal`, `low`, `medium`, and `high`. Reducing
-    # reasoning effort can result in faster responses and fewer tokens used on
-    # reasoning in a response.
+    # supported values are `none`, `minimal`, `low`, `medium`, `high`, and `xhigh`.
+    # Reducing reasoning effort can result in faster responses and fewer tokens used
+    # on reasoning in a response.
     #
     # - `gpt-5.1` defaults to `none`, which does not perform reasoning. The supported
     #   reasoning values for `gpt-5.1` are `none`, `low`, `medium`, and `high`. Tool
@@ -14,6 +14,7 @@ module OpenAI
     # - All models before `gpt-5.1` default to `medium` reasoning effort, and do not
     #   support `none`.
     # - The `gpt-5-pro` model defaults to (and only supports) `high` reasoning effort.
+    # - `xhigh` is currently only supported for `gpt-5.1-codex-max`.
     module ReasoningEffort
       extend OpenAI::Internal::Type::Enum
 
@@ -22,6 +23,7 @@ module OpenAI
       LOW = :low
       MEDIUM = :medium
       HIGH = :high
+      XHIGH = :xhigh
 
       # @!method self.values
       #   @return [Array<Symbol>]

data/lib/openai/models/responses/compacted_response.rb

@@ -0,0 +1,56 @@
+# frozen_string_literal: true
+
+module OpenAI
+  module Models
+    module Responses
+      # @see OpenAI::Resources::Responses#compact
+      class CompactedResponse < OpenAI::Internal::Type::BaseModel
+        # @!attribute id
+        #   The unique identifier for the compacted response.
+        #
+        #   @return [String]
+        required :id, String
+
+        # @!attribute created_at
+        #   Unix timestamp (in seconds) when the compacted conversation was created.
+        #
+        #   @return [Integer]
+        required :created_at, Integer
+
+        # @!attribute object
+        #   The object type. Always `response.compaction`.
+        #
+        #   @return [Symbol, :"response.compaction"]
+        required :object, const: :"response.compaction"
+
+        # @!attribute output
+        #   The compacted list of output items. This is a list of all user messages,
+        #   followed by a single compaction item.
+        #
+        #   @return [Array<OpenAI::Models::Responses::ResponseOutputMessage, OpenAI::Models::Responses::ResponseFileSearchToolCall, OpenAI::Models::Responses::ResponseFunctionToolCall, OpenAI::Models::Responses::ResponseFunctionWebSearch, OpenAI::Models::Responses::ResponseComputerToolCall, OpenAI::Models::Responses::ResponseReasoningItem, OpenAI::Models::Responses::ResponseCompactionItem, OpenAI::Models::Responses::ResponseOutputItem::ImageGenerationCall, OpenAI::Models::Responses::ResponseCodeInterpreterToolCall, OpenAI::Models::Responses::ResponseOutputItem::LocalShellCall, OpenAI::Models::Responses::ResponseFunctionShellToolCall, OpenAI::Models::Responses::ResponseFunctionShellToolCallOutput, OpenAI::Models::Responses::ResponseApplyPatchToolCall, OpenAI::Models::Responses::ResponseApplyPatchToolCallOutput, OpenAI::Models::Responses::ResponseOutputItem::McpCall, OpenAI::Models::Responses::ResponseOutputItem::McpListTools, OpenAI::Models::Responses::ResponseOutputItem::McpApprovalRequest, OpenAI::Models::Responses::ResponseCustomToolCall>]
+        required :output, -> { OpenAI::Internal::Type::ArrayOf[union: OpenAI::Responses::ResponseOutputItem] }
+
+        # @!attribute usage
+        #   Token accounting for the compaction pass, including cached, reasoning, and total
+        #   tokens.
+        #
+        #   @return [OpenAI::Models::Responses::ResponseUsage]
+        required :usage, -> { OpenAI::Responses::ResponseUsage }
+
+        # @!method initialize(id:, created_at:, output:, usage:, object: :"response.compaction")
+        #   Some parameter documentations has been truncated, see
+        #   {OpenAI::Models::Responses::CompactedResponse} for more details.
+        #
+        #   @param id [String] The unique identifier for the compacted response.
+        #
+        #   @param created_at [Integer] Unix timestamp (in seconds) when the compacted conversation was created.
+        #
+        #   @param output [Array<OpenAI::Models::Responses::ResponseOutputMessage, OpenAI::Models::Responses::ResponseFileSearchToolCall, OpenAI::Models::Responses::ResponseFunctionToolCall, OpenAI::Models::Responses::ResponseFunctionWebSearch, OpenAI::Models::Responses::ResponseComputerToolCall, OpenAI::Models::Responses::ResponseReasoningItem, OpenAI::Models::Responses::ResponseCompactionItem, OpenAI::Models::Responses::ResponseOutputItem::ImageGenerationCall, OpenAI::Models::Responses::ResponseCodeInterpreterToolCall, OpenAI::Models::Responses::ResponseOutputItem::LocalShellCall, OpenAI::Models::Responses::ResponseFunctionShellToolCall, OpenAI::Models::Responses::ResponseFunctionShellToolCallOutput, OpenAI::Models::Responses::ResponseApplyPatchToolCall, OpenAI::Models::Responses::ResponseApplyPatchToolCallOutput, OpenAI::Models::Responses::ResponseOutputItem::McpCall, OpenAI::Models::Responses::ResponseOutputItem::McpListTools, OpenAI::Models::Responses::ResponseOutputItem::McpApprovalRequest, OpenAI::Models::Responses::ResponseCustomToolCall>] The compacted list of output items. This is a list of all user messages, followe
+        #
+        #   @param usage [OpenAI::Models::Responses::ResponseUsage] Token accounting for the compaction pass, including cached, reasoning, and total
+        #
+        #   @param object [Symbol, :"response.compaction"] The object type. Always `response.compaction`.
+      end
+    end
+  end
+end