diff --git a/src/backend/drivers/ai-speech2txt/SpeechToTextDriver.test.ts b/src/backend/drivers/ai-speech2txt/SpeechToTextDriver.test.ts
index a0e1249b4..946d94250 100644
--- a/src/backend/drivers/ai-speech2txt/SpeechToTextDriver.test.ts
+++ b/src/backend/drivers/ai-speech2txt/SpeechToTextDriver.test.ts
@@ -531,7 +531,6 @@ describe('SpeechToTextDriver model selection and payload shape', () => {
                 file: dataUrl(Buffer.from('a'), 'audio/mp3'),
                 chunking_strategy: 'auto',
                 known_speaker_names: ['agent'],
-                known_speaker_references: ['data:audio/wav;base64,QQ=='],
             }),
         );
 
@@ -539,7 +538,6 @@ describe('SpeechToTextDriver model selection and payload shape', () => {
         const sent = transcriptionsCreateMock.mock.calls[0]![0];
         expect(sent.chunking_strategy).toBeUndefined();
         expect(sent.known_speaker_names).toBeUndefined();
-        expect(sent.known_speaker_references).toBeUndefined();
     });
 
     it('passes extra_body through verbatim', async () => {
diff --git a/src/backend/drivers/ai-speech2txt/providers/xai/XAISpeechToTextProvider.test.ts b/src/backend/drivers/ai-speech2txt/providers/xai/XAISpeechToTextProvider.test.ts
index 07b07753d..0241aa3ba 100644
--- a/src/backend/drivers/ai-speech2txt/providers/xai/XAISpeechToTextProvider.test.ts
+++ b/src/backend/drivers/ai-speech2txt/providers/xai/XAISpeechToTextProvider.test.ts
@@ -453,7 +453,6 @@ describe('XAISpeechToTextProvider request shape', () => {
                 timestamp_granularities: ['word'],
                 chunking_strategy: 'auto',
                 known_speaker_names: ['agent'],
-                known_speaker_references: ['data:audio/wav;base64,QQ=='],
             }),
         );
 
@@ -465,8 +464,6 @@ describe('XAISpeechToTextProvider request shape', () => {
         expect(form.get('chunking_strategy')).toBeNull();
         expect(form.get('known_speaker_names')).toBeNull();
         expect(form.get('known_speaker_names[]')).toBeNull();
-        expect(form.get('known_speaker_references')).toBeNull();
-        expect(form.get('known_speaker_references[]')).toBeNull();
     });
 
     it('forwards a pinned grok-voice-transcribe version as `model`', async () => {
diff --git a/src/backend/drivers/ai-speech2txt/types.ts b/src/backend/drivers/ai-speech2txt/types.ts
index 8432797c4..73ab8f2aa 100644
--- a/src/backend/drivers/ai-speech2txt/types.ts
+++ b/src/backend/drivers/ai-speech2txt/types.ts
@@ -53,7 +53,6 @@ export interface ITranscribeArgs {
     timestamp_granularities?: string[];
     chunking_strategy?: string;
     known_speaker_names?: string[];
-    known_speaker_references?: unknown[];
     extra_body?: Record<string, unknown>;
     stream?: boolean;
     test_mode?: boolean;
diff --git a/src/docs/src/AI/speech2txt.md b/src/docs/src/AI/speech2txt.md
index 4ad0105c7..9f0c65ec6 100644
--- a/src/docs/src/AI/speech2txt.md
+++ b/src/docs/src/AI/speech2txt.md
@@ -44,8 +44,8 @@ Fine-tune how transcription runs.
 - `stream` (Boolean): Reserved for future streaming support. Streaming is not currently supported.
 - `test_mode` (Boolean): When `true`, returns a sample response without using credits. Defaults to `false`.
 
-The options `timestamp_granularities`, `chunking_strategy`, `known_speaker_names`,
-and `known_speaker_references` are accepted and ignored by both providers.
+The options `timestamp_granularities`, `chunking_strategy`, and `known_speaker_names`
+are accepted and ignored by both providers.
 
 **xAI-specific options** (when `provider: 'xai'`):
 
diff --git a/src/puter-js/src/modules/ai/ai.test.js b/src/puter-js/src/modules/ai/ai.test.js
index 31f5ae47f..365f65c6d 100644
--- a/src/puter-js/src/modules/ai/ai.test.js
+++ b/src/puter-js/src/modules/ai/ai.test.js
@@ -567,7 +567,6 @@ describe('ai.speech2txt driver payloads', () => {
             timestamp_granularities: ['word'],
             chunking_strategy: 'auto',
             known_speaker_names: ['agent'],
-            known_speaker_references: ['data:audio/wav;base64,QQ=='],
         };
         await ai.speech2txt({ file: 'data:audio/mpeg;base64,QUJD', ...legacy });
         expect(lastBody().args).toEqual({
diff --git a/src/puter-js/src/modules/ai/types.js b/src/puter-js/src/modules/ai/types.js
index 16d34a2bc..ebe312a2f 100644
--- a/src/puter-js/src/modules/ai/types.js
+++ b/src/puter-js/src/modules/ai/types.js
@@ -483,7 +483,6 @@
  * @property {string[]} [timestamp_granularities] Accepted and ignored by all providers.
  * @property {string} [chunking_strategy] Accepted and ignored by all providers.
  * @property {string[]} [known_speaker_names] Accepted and ignored by all providers.
- * @property {string[]} [known_speaker_references] Accepted and ignored by all providers.
  * @property {Record<string, unknown>} [extra_body]
  * @property {boolean} [format]
  * @property {boolean} [diarize]
diff --git a/src/puter-js/tests/api/suites/ai.suite.ts b/src/puter-js/tests/api/suites/ai.suite.ts
index 9c2d454c8..4cd26d668 100644
--- a/src/puter-js/tests/api/suites/ai.suite.ts
+++ b/src/puter-js/tests/api/suites/ai.suite.ts
@@ -1307,7 +1307,6 @@ export default suite('ai', {
             timestamp_granularities: ['word'],
             chunking_strategy: 'auto',
             known_speaker_names: ['speaker'],
-            known_speaker_references: [TINY_AUDIO],
             test_mode: true,
         })) as { text?: string };
         t.assert.ok((result.text ?? '').includes('sample transcription'));