feat(api): realtime 2

Author: stainless-app[bot]

.stats.yml

@@ -1,4 +1,4 @@
 configured_endpoints: 232
-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/resources/realtime/api.md

@@ -56,6 +56,8 @@ Types:
 - <code><a href="./src/resources/realtime/realtime.ts">RealtimeMcpToolCall</a></code>
 - <code><a href="./src/resources/realtime/realtime.ts">RealtimeMcpToolExecutionError</a></code>
 - <code><a href="./src/resources/realtime/realtime.ts">RealtimeMcphttpError</a></code>
+- <code><a href="./src/resources/realtime/realtime.ts">RealtimeReasoning</a></code>
+- <code><a href="./src/resources/realtime/realtime.ts">RealtimeReasoningEffort</a></code>
 - <code><a href="./src/resources/realtime/realtime.ts">RealtimeResponse</a></code>
 - <code><a href="./src/resources/realtime/realtime.ts">RealtimeResponseCreateAudioOutput</a></code>
 - <code><a href="./src/resources/realtime/realtime.ts">RealtimeResponseCreateMcpTool</a></code>
@@ -124,7 +126,6 @@ Types:
 
 Types:
 
-- <code><a href="./src/resources/realtime/client-secrets.ts">RealtimeSessionClientSecret</a></code>
 - <code><a href="./src/resources/realtime/client-secrets.ts">RealtimeSessionCreateResponse</a></code>
 - <code><a href="./src/resources/realtime/client-secrets.ts">RealtimeTranscriptionSessionCreateResponse</a></code>
 - <code><a href="./src/resources/realtime/client-secrets.ts">RealtimeTranscriptionSessionTurnDetection</a></code>

src/resources/realtime/calls.ts

@@ -134,6 +134,7 @@ export interface CallAcceptParams {
     | (string & {})
     | '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'
@@ -157,12 +158,23 @@ export interface CallAcceptParams {
    */
   output_modalities?: Array<'text' | 'audio'>;
 
+  /**
+   * Whether the model may call multiple tools in parallel. Only supported by
+   * reasoning Realtime models such as `gpt-realtime-2`.
+   */
+  parallel_tool_calls?: boolean;
+
   /**
    * Reference to a prompt template and its variables.
    * [Learn more](https://platform.openai.com/docs/guides/text?api-mode=responses#reusable-prompts).
    */
   prompt?: ResponsesAPI.ResponsePrompt | null;
 
+  /**
+   * Configuration for reasoning-capable Realtime models such as `gpt-realtime-2`.
+   */
+  reasoning?: RealtimeAPI.RealtimeReasoning;
+
   /**
    * How the model chooses tools. Provide one of the string modes or force a specific
    * function/MCP tool.

src/resources/realtime/client-secrets.ts

@@ -41,32 +41,18 @@ export class ClientSecrets extends APIResource {
 }
 
 /**
- * Ephemeral key returned by the API.
+ * A Realtime session configuration object.
  */
-export interface RealtimeSessionClientSecret {
-  /**
-   * Timestamp for when the token expires. Currently, all tokens expire after one
-   * minute.
-   */
-  expires_at: number;
-
+export interface RealtimeSessionCreateResponse {
   /**
-   * Ephemeral key usable in client environments to authenticate connections to the
-   * Realtime API. Use this in client-side environments rather than a standard API
-   * token, which should only be used server-side.
+   * Unique identifier for the session that looks like `sess_1234567890abcdef`.
    */
-  value: string;
-}
+  id: string;
 
-/**
- * A new Realtime session configuration, with an ephemeral key. Default TTL for
- * keys is one minute.
- */
-export interface RealtimeSessionCreateResponse {
   /**
-   * Ephemeral key returned by the API.
+   * The object type. Always `realtime.session`.
    */
-  client_secret: RealtimeSessionClientSecret;
+  object: 'realtime.session';
 
   /**
    * The type of session to create. Always `realtime` for the Realtime API.
@@ -78,6 +64,11 @@ export interface RealtimeSessionCreateResponse {
    */
   audio?: RealtimeSessionCreateResponse.Audio;
 
+  /**
+   * Expiration timestamp for the session, in seconds since epoch.
+   */
+  expires_at?: number;
+
   /**
    * Additional fields to include in server outputs.
    *
@@ -115,6 +106,7 @@ export interface RealtimeSessionCreateResponse {
     | (string & {})
     | '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'
@@ -144,6 +136,11 @@ export interface RealtimeSessionCreateResponse {
    */
   prompt?: ResponsesAPI.ResponsePrompt | null;
 
+  /**
+   * Configuration for reasoning-capable Realtime models such as `gpt-realtime-2`.
+   */
+  reasoning?: RealtimeAPI.RealtimeReasoning;
+
   /**
    * How the model chooses tools. Provide one of the string modes or force a specific
    * function/MCP tool.
@@ -215,16 +212,6 @@ export namespace RealtimeSessionCreateResponse {
        */
       noise_reduction?: Input.NoiseReduction;
 
-      /**
-       * Configuration for input audio transcription, defaults to off and can be set to
-       * `null` to turn off once on. Input audio transcription is not native to the
-       * model, since the model consumes audio directly. Transcription runs
-       * asynchronously through
-       * [the /audio/transcriptions endpoint](https://platform.openai.com/docs/api-reference/audio/createTranscription)
-       * and should be treated as guidance of input audio content rather than precisely
-       * what the model heard. The client can optionally set the language and prompt for
-       * transcription, these offer additional guidance to the transcription service.
-       */
       transcription?: RealtimeAPI.AudioTranscription;
 
       /**
@@ -241,6 +228,9 @@ export namespace RealtimeSessionCreateResponse {
        * 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.
        */
       turn_detection?: Input.ServerVad | Input.SemanticVad | null;
     }
@@ -640,17 +630,15 @@ export namespace RealtimeTranscriptionSessionCreateResponse {
        */
       noise_reduction?: Input.NoiseReduction;
 
-      /**
-       * Configuration of the transcription model.
-       */
       transcription?: RealtimeAPI.AudioTranscription;
 
       /**
        * Configuration for turn detection. Can be set to `null` to turn off. Server VAD
        * means that the model will detect the start and end of speech based on audio
-       * volume and respond at the end of user speech.
+       * volume and respond at the end of user speech. For `gpt-realtime-whisper`, this
+       * must be `null`; VAD is not supported.
        */
-      turn_detection?: ClientSecretsAPI.RealtimeTranscriptionSessionTurnDetection;
+      turn_detection?: ClientSecretsAPI.RealtimeTranscriptionSessionTurnDetection | null;
     }
 
     export namespace Input {
@@ -672,7 +660,8 @@ export namespace RealtimeTranscriptionSessionCreateResponse {
 /**
  * Configuration for turn detection. Can be set to `null` to turn off. Server VAD
  * means that the model will detect the start and end of speech based on audio
- * volume and respond at the end of user speech.
+ * volume and respond at the end of user speech. For `gpt-realtime-whisper`, this
+ * must be `null`; VAD is not supported.
  */
 export interface RealtimeTranscriptionSessionTurnDetection {
   /**
@@ -763,7 +752,6 @@ export namespace ClientSecretCreateParams {
 
 export declare namespace ClientSecrets {
   export {
-    type RealtimeSessionClientSecret as RealtimeSessionClientSecret,
     type RealtimeSessionCreateResponse as RealtimeSessionCreateResponse,
     type RealtimeTranscriptionSessionCreateResponse as RealtimeTranscriptionSessionCreateResponse,
     type RealtimeTranscriptionSessionTurnDetection as RealtimeTranscriptionSessionTurnDetection,

src/resources/realtime/index.ts

@@ -3,7 +3,6 @@
 export { Calls, type CallAcceptParams, type CallReferParams, type CallRejectParams } from './calls';
 export {
   ClientSecrets,
-  type RealtimeSessionClientSecret,
   type RealtimeSessionCreateResponse,
   type RealtimeTranscriptionSessionCreateResponse,
   type RealtimeTranscriptionSessionTurnDetection,