diff --git a/litellm/litellm_core_utils/prompt_templates/factory.py b/litellm/litellm_core_utils/prompt_templates/factory.py
index 7b485501f61..88dd9b57584 100644
--- a/litellm/litellm_core_utils/prompt_templates/factory.py
+++ b/litellm/litellm_core_utils/prompt_templates/factory.py
@@ -4131,6 +4131,7 @@ async def _bedrock_converse_messages_pt_async(  # noqa: PLR0915
                                         OpenAIMessageContentListBlock, element
                                     ),
                                     block_type="content_block",
+                                    model=model,
                                 )
                             )
                             if _cache_point_block is not None:
@@ -4142,7 +4143,7 @@ async def _bedrock_converse_messages_pt_async(  # noqa: PLR0915
                     _part = BedrockContentBlock(text=messages[msg_i]["content"])
                     _cache_point_block = (
                         litellm.AmazonConverseConfig()._get_cache_point_block(
-                            message_block, block_type="content_block"
+                            message_block, block_type="content_block", model=model
                         )
                     )
                     user_content.append(_part)
@@ -4294,6 +4295,7 @@ async def _bedrock_converse_messages_pt_async(  # noqa: PLR0915
                                     OpenAIMessageContentListBlock, element
                                 ),
                                 block_type="content_block",
+                                model=model,
                             )
                         )
                         if _cache_point_block is not None:
@@ -4311,7 +4313,7 @@ async def _bedrock_converse_messages_pt_async(  # noqa: PLR0915
                     # Add cache point block for assistant string content
                     _cache_point_block = (
                         litellm.AmazonConverseConfig()._get_cache_point_block(
-                            assistant_message_block, block_type="content_block"
+                            assistant_message_block, block_type="content_block", model=model
                         )
                     )
                     if _cache_point_block is not None:
@@ -4505,6 +4507,7 @@ def _bedrock_converse_messages_pt(  # noqa: PLR0915
                                     OpenAIMessageContentListBlock, element
                                 ),
                                 block_type="content_block",
+                                model=model,
                             )
                         )
                         if _cache_point_block is not None:
@@ -4514,7 +4517,7 @@ def _bedrock_converse_messages_pt(  # noqa: PLR0915
                 _part = BedrockContentBlock(text=messages[msg_i]["content"])
                 _cache_point_block = (
                     litellm.AmazonConverseConfig()._get_cache_point_block(
-                        message_block, block_type="content_block"
+                        message_block, block_type="content_block", model=model
                     )
                 )
                 user_content.append(_part)
@@ -4659,6 +4662,7 @@ def _bedrock_converse_messages_pt(  # noqa: PLR0915
                                     OpenAIMessageContentListBlock, element
                                 ),
                                 block_type="content_block",
+                                model=model,
                             )
                         )
                         if _cache_point_block is not None:
@@ -4671,7 +4675,7 @@ def _bedrock_converse_messages_pt(  # noqa: PLR0915
                 # Add cache point block for assistant string content
                 _cache_point_block = (
                     litellm.AmazonConverseConfig()._get_cache_point_block(
-                        assistant_message_block, block_type="content_block"
+                        assistant_message_block, block_type="content_block", model=model
                     )
                 )
                 if _cache_point_block is not None:
diff --git a/tests/test_litellm/llms/bedrock/chat/test_converse_transformation.py b/tests/test_litellm/llms/bedrock/chat/test_converse_transformation.py
index 26395597166..ef2c63f7fa1 100644
--- a/tests/test_litellm/llms/bedrock/chat/test_converse_transformation.py
+++ b/tests/test_litellm/llms/bedrock/chat/test_converse_transformation.py
@@ -1318,6 +1318,47 @@ async def test_assistant_message_cache_control():
     assert assistant_content[1]["cachePoint"]["type"] == "default"
 
 
+@pytest.mark.asyncio
+async def test_assistant_message_cache_control_with_ttl_claude_45():
+    """Test cache_control TTL is preserved for Claude 4.5 Bedrock models."""
+    from litellm.litellm_core_utils.prompt_templates.factory import (
+        BedrockConverseMessagesProcessor,
+        _bedrock_converse_messages_pt,
+    )
+
+    messages = [
+        {"role": "user", "content": "Hello"},
+        {
+            "role": "assistant",
+            "content": "Hi there!",
+            "cache_control": {"type": "ephemeral", "ttl": "1h"},
+        },
+    ]
+
+    model = "bedrock/anthropic.claude-sonnet-4.5-20250514-v1:0"
+
+    result = _bedrock_converse_messages_pt(
+        messages=messages,
+        model=model,
+        llm_provider="bedrock_converse",
+    )
+
+    async_result = (
+        await BedrockConverseMessagesProcessor._bedrock_converse_messages_pt_async(
+            messages=messages,
+            model=model,
+            llm_provider="bedrock_converse",
+        )
+    )
+
+    assert result == async_result
+
+    assistant_content = result[1]["content"]
+    assert "cachePoint" in assistant_content[1]
+    assert assistant_content[1]["cachePoint"]["type"] == "default"
+    assert assistant_content[1]["cachePoint"]["ttl"] == "1h"
+
+
 @pytest.mark.asyncio
 async def test_assistant_message_list_content_cache_control():
     """Test assistant messages with list content and cache_control."""