diff --git a/extensions/openai/realtime-voice-provider.test.ts b/extensions/openai/realtime-voice-provider.test.ts
index a5948d42f493..0cc9cba4a309 100644
--- a/extensions/openai/realtime-voice-provider.test.ts
+++ b/extensions/openai/realtime-voice-provider.test.ts
@@ -289,7 +289,7 @@ describe("buildOpenAIRealtimeVoiceProvider", () => {
   it("declares realtime Talk capabilities for catalog selection", () => {
     const provider = buildOpenAIRealtimeVoiceProvider();
 
-    expect(provider.defaultModel).toBe("gpt-realtime-2");
+    expect(provider.defaultModel).toBe("gpt-realtime-2.1");
     expect(provider.capabilities).toEqual({
       transports: ["webrtc", "gateway-relay"],
       inputAudioFormats: [
@@ -541,7 +541,7 @@ describe("buildOpenAIRealtimeVoiceProvider", () => {
     });
     const body = requireFetchJsonBody();
     const bodySession = requireRecord(body.session, "fetch session");
-    expect(bodySession.model).toBe("gpt-realtime-2");
+    expect(bodySession.model).toBe("gpt-realtime-2.1");
     expect(requireNestedRecord(bodySession, ["audio", "input"])).toEqual({
       noise_reduction: { type: "near_field" },
       turn_detection: {
@@ -558,7 +558,7 @@ describe("buildOpenAIRealtimeVoiceProvider", () => {
       transport: "webrtc",
       clientSecret: "client-secret-123",
       offerUrl: "https://api.openai.com/v1/realtime/calls",
-      model: "gpt-realtime-2",
+      model: "gpt-realtime-2.1",
       expiresAt: 1_765_000_000_000,
     });
     // originator, version, and User-Agent are server-side attribution headers; they
@@ -920,7 +920,7 @@ describe("buildOpenAIRealtimeVoiceProvider", () => {
     const session = requireSession(socket);
     expectRecordFields(session, "session", {
       type: "realtime",
-      model: "gpt-realtime-2",
+      model: "gpt-realtime-2.1",
       output_modalities: ["audio"],
     });
     const inputAudio = requireNestedRecord(session, ["audio", "input"]);
diff --git a/extensions/openai/realtime-voice-provider.ts b/extensions/openai/realtime-voice-provider.ts
index a0ebe98f49f9..d84cef4fa571 100644
--- a/extensions/openai/realtime-voice-provider.ts
+++ b/extensions/openai/realtime-voice-provider.ts
@@ -86,7 +86,7 @@ type OpenAIRealtimeVoiceBridgeConfig = RealtimeVoiceBridgeCreateRequest & {
   azureApiVersion?: string;
 };
 
-const OPENAI_REALTIME_DEFAULT_MODEL = "gpt-realtime-2";
+const OPENAI_REALTIME_DEFAULT_MODEL = "gpt-realtime-2.1";
 const OPENAI_REALTIME_INPUT_TRANSCRIPTION_MODEL = "gpt-4o-mini-transcribe";
 const OPENAI_REALTIME_ACTIVE_RESPONSE_ERROR_PREFIX =
   "Conversation already has an active response in progress:";
diff --git a/scripts/dev/realtime-talk-live-smoke.ts b/scripts/dev/realtime-talk-live-smoke.ts
index 8a276af7daa1..554edddd5ab6 100644
--- a/scripts/dev/realtime-talk-live-smoke.ts
+++ b/scripts/dev/realtime-talk-live-smoke.ts
@@ -11,7 +11,7 @@ import {
 } from "../lib/dev-tooling-safety.ts";
 
 const OPENAI_REALTIME_MODEL =
-  process.env.OPENCLAW_REALTIME_OPENAI_MODEL?.trim() || "gpt-realtime-2";
+  process.env.OPENCLAW_REALTIME_OPENAI_MODEL?.trim() || "gpt-realtime-2.1";
 const OPENAI_REALTIME_VOICE = process.env.OPENCLAW_REALTIME_OPENAI_VOICE?.trim() || "alloy";
 const DEFAULT_OPENAI_HTTP_TIMEOUT_MS = 30_000;
 const OPENAI_HTTP_RESPONSE_MAX_BYTES = 256 * 1024;