diff --git a/src/pipecat/frames/frames.py b/src/pipecat/frames/frames.py
index 90abfeda6..320d25e13 100644
--- a/src/pipecat/frames/frames.py
+++ b/src/pipecat/frames/frames.py
@@ -186,6 +186,13 @@ class LLMSetToolsFrame(DataFrame):
     tools: List[dict]
 
 
+@dataclass
+class LLMEnablePromptCachingFrame(DataFrame):
+    """A frame to enable/disable prompt caching in certain LLMs.
+    """
+    enable: bool
+
+
 @dataclass
 class TTSSpeakFrame(DataFrame):
     """A frame that contains a text that should be spoken by the TTS in the
diff --git a/src/pipecat/services/anthropic.py b/src/pipecat/services/anthropic.py
index 088db7861..bfe2e86f1 100644
--- a/src/pipecat/services/anthropic.py
+++ b/src/pipecat/services/anthropic.py
@@ -16,6 +16,7 @@ import re
 
 from pipecat.frames.frames import (
     Frame,
+    LLMEnablePromptCachingFrame,
     LLMModelUpdateFrame,
     TextFrame,
     VisionImageRawFrame,
@@ -62,10 +63,10 @@ class AnthropicContextAggregatorPair:
     _user: 'AnthropicUserContextAggregator'
     _assistant: 'AnthropicAssistantContextAggregator'
 
-    def user(self) -> str:
+    def user(self) -> 'AnthropicUserContextAggregator':
         return self._user
 
-    def assistant(self) -> str:
+    def assistant(self) -> 'AnthropicAssistantContextAggregator':
         return self._assistant
 
 
@@ -227,6 +228,9 @@ class AnthropicLLMService(LLMService):
         elif isinstance(frame, LLMModelUpdateFrame):
             logger.debug(f"Switching LLM model to: [{frame.model}]")
             self._model = frame.model
+        elif isinstance(frame, LLMEnablePromptCachingFrame):
+            logger.debug(f"Setting enable prompt caching to: [{frame.enable}]")
+            self._enable_prompt_caching_beta = frame.enable
         else:
             await self.push_frame(frame, direction)
 
diff --git a/src/pipecat/services/openai.py b/src/pipecat/services/openai.py
index 12708647c..52da196d2 100644
--- a/src/pipecat/services/openai.py
+++ b/src/pipecat/services/openai.py
@@ -229,10 +229,10 @@ class OpenAIContextAggregatorPair:
     _user: 'OpenAIUserContextAggregator'
     _assistant: 'OpenAIAssistantContextAggregator'
 
-    def user(self) -> str:
+    def user(self) -> 'OpenAIUserContextAggregator':
         return self._user
 
-    def assistant(self) -> str:
+    def assistant(self) -> 'OpenAIAssistantContextAggregator':
         return self._assistant
 
 
diff --git a/src/pipecat/services/together.py b/src/pipecat/services/together.py
index bf34dd099..685609858 100644
--- a/src/pipecat/services/together.py
+++ b/src/pipecat/services/together.py
@@ -49,10 +49,10 @@ class TogetherContextAggregatorPair:
     _user: 'TogetherUserContextAggregator'
     _assistant: 'TogetherAssistantContextAggregator'
 
-    def user(self) -> str:
+    def user(self) -> 'TogetherUserContextAggregator':
         return self._user
 
-    def assistant(self) -> str:
+    def assistant(self) -> 'TogetherAssistantContextAggregator':
         return self._assistant
 
 
@@ -75,7 +75,7 @@ class TogetherLLMService(LLMService):
     def can_generate_metrics(self) -> bool:
         return True
 
-    @ staticmethod
+    @staticmethod
     def create_context_aggregator(context: OpenAILLMContext) -> TogetherContextAggregatorPair:
         user = TogetherUserContextAggregator(context)
         assistant = TogetherAssistantContextAggregator(user)
@@ -191,14 +191,14 @@ class TogetherLLMContext(OpenAILLMContext):
     ):
         super().__init__(messages=messages)
 
-    @ classmethod
+    @classmethod
     def from_openai_context(cls, openai_context: OpenAILLMContext):
         self = cls(
             messages=openai_context.messages,
         )
         return self
 
-    @ classmethod
+    @classmethod
     def from_messages(cls, messages: List[dict]) -> "TogetherLLMContext":
         return cls(messages=messages)