diff --git a/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py b/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py
index 4cf6b33cd10..8c3513e064e 100644
--- a/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py
+++ b/litellm/llms/vertex_ai/gemini/vertex_and_google_ai_studio_gemini.py
@@ -147,9 +147,7 @@ _GEMINI_SPEECH_CONFIG_KEY_MAP: Final[Mapping[str, str]] = MappingProxyType(
 
 
 def _normalize_gemini_speech_config_item(value: Mapping[str, object]) -> dict[str, object]:
-    normalized: Final[
-        dict[str, object]
-    ] = {}  # mutable-ok: iterative traversal preserves deeply nested provider payloads
+    normalized: Final[dict[str, object]] = {}
 
     pending: Final[  # mutable-ok: the explicit worklist must grow while avoiding recursive traversal
         list[tuple[dict[str, object] | list[object], Mapping[object, object] | list[object]]]
@@ -167,9 +165,9 @@ def _normalize_gemini_speech_config_item(value: Mapping[str, object]) -> dict[st
 
             child: object
             if isinstance(item, Mapping):
-                child = {}  # mutable-ok: nested mappings are filled by the worklist
+                child = {}
             elif isinstance(item, list):
-                child = []  # mutable-ok: nested lists are filled by the worklist
+                child = []
             else:
                 child = item
 
@@ -207,7 +205,7 @@ def normalize_gemini_speech_config(
         return {key: item for key, item in normalized.items() if key not in ("name", "voiceName", "voice")}
     if voice_name is None:
         return normalized
-    return {  # mutable-ok: provider request serialization requires a concrete dict
+    return {
         **{key: item for key, item in normalized.items() if key not in ("name", "voiceName", "voice", "modelName")},
         "voiceConfig": {"prebuiltVoiceConfig": {"voiceName": voice_name}},
     }
@@ -1136,19 +1134,19 @@ class VertexGeminiConfig(VertexAIBaseConfig, BaseConfig):
         speech_config_without_language: Final[  # mutable-ok: provider request serialization requires a concrete dict
             dict[str, object]
         ] = (
-            {  # mutable-ok: provider request serialization requires a concrete dict
-                "voiceConfig": {  # mutable-ok: nested provider payload is serialized as a dict
-                    "prebuiltVoiceConfig": {  # mutable-ok: nested provider payload is serialized as a dict
+            {
+                "voiceConfig": {
+                    "prebuiltVoiceConfig": {
                         "voiceName": value["voice"],
                     }
                 }
             }
             if "voice" in value
-            else {}  # mutable-ok: provider request serialization requires a concrete empty dict
+            else {}
         )
 
         if language_code is not None:
-            return {  # mutable-ok: provider request serialization requires a concrete dict
+            return {
                 **speech_config_without_language,
                 "languageCode": language_code,
             }
diff --git a/litellm/llms/vertex_ai/text_to_speech/transformation.py b/litellm/llms/vertex_ai/text_to_speech/transformation.py
index 3fcbcb898c0..261c5f09768 100644
--- a/litellm/llms/vertex_ai/text_to_speech/transformation.py
+++ b/litellm/llms/vertex_ai/text_to_speech/transformation.py
@@ -230,16 +230,16 @@ class VertexAITextToSpeechConfig(BaseTextToSpeechConfig, VertexBase):
             "multi_speaker_voice_config",
         )
         if multi_speaker_config is None:
-            return []  # mutable-ok: provider request serialization requires a concrete empty list
+            return []
         raw_speaker_configs: Final = multi_speaker_config.get(
             "speakerVoiceConfigs",
             multi_speaker_config.get(
                 "speaker_voice_configs",
-                [],  # mutable-ok: missing speaker configuration uses a concrete empty-list sentinel
+                [],
             ),
         )
         if not isinstance(raw_speaker_configs, list):
-            return []  # mutable-ok: malformed speaker configuration produces a concrete empty list
+            return []
         speaker_configs: Final[  # mutable-ok: validated payloads are accumulated for serialization
             list[VertexTextToSpeechSpeakerVoiceConfig]
         ] = []
@@ -269,7 +269,7 @@ class VertexAITextToSpeechConfig(BaseTextToSpeechConfig, VertexBase):
                         )
             if speaker_alias is not None and speaker_id is not None:
                 speaker_configs.append(
-                    {  # mutable-ok: provider request serialization requires a concrete dict
+                    {
                         "speakerAlias": speaker_alias,
                         "speakerId": speaker_id,
                     }
@@ -305,7 +305,7 @@ class VertexAITextToSpeechConfig(BaseTextToSpeechConfig, VertexBase):
         voice: str | Mapping[str, object],
     ) -> tuple[str | None, dict[str, object]]:  # mutable-ok: provider request serialization requires a concrete dict
         if isinstance(voice, str):
-            return voice, {  # mutable-ok: provider request serialization requires a concrete dict
+            return voice, {
                 "languageCode": self.DEFAULT_LANGUAGE_CODE,
                 "modelName": model,
                 "name": voice,
@@ -320,21 +320,21 @@ class VertexAITextToSpeechConfig(BaseTextToSpeechConfig, VertexBase):
         model_name: Final = model
         speaker_configs: Final = self._extract_gemini_tts_speaker_configs(voice)
         if speaker_configs:
-            return None, {  # mutable-ok: provider request serialization requires a concrete dict
+            return None, {
                 "languageCode": language_code,
                 "modelName": model_name,
-                "multiSpeakerVoiceConfig": {  # mutable-ok: nested provider payload is serialized as a dict
+                "multiSpeakerVoiceConfig": {
                     "speakerVoiceConfigs": speaker_configs,
                 },
             }
         voice_name: Final = self._extract_gemini_tts_voice_name(voice)
         if voice_name is not None:
-            return None, {  # mutable-ok: provider request serialization requires a concrete dict
+            return None, {
                 "languageCode": language_code,
                 "modelName": model_name,
                 "name": voice_name,
             }
-        return None, {  # mutable-ok: provider request serialization requires a concrete dict
+        return None, {
             **voice,
             "languageCode": language_code,
             "modelName": model_name,