diff --git a/src/strands/models/bedrock.py b/src/strands/models/bedrock.py
index 8a6d5116f..aa19b114d 100644
--- a/src/strands/models/bedrock.py
+++ b/src/strands/models/bedrock.py
@@ -293,7 +293,9 @@ def _format_bedrock_messages(self, messages: Messages) -> Messages:
             https://docs.aws.amazon.com/bedrock/latest/APIReference/API_runtime_ToolResultBlock.html
         """
         cleaned_messages = []
+
         filtered_unknown_members = False
+        dropped_deepseek_reasoning_content = False
 
         for message in messages:
             cleaned_content: list[ContentBlock] = []
@@ -304,6 +306,12 @@ def _format_bedrock_messages(self, messages: Messages) -> Messages:
                     filtered_unknown_members = True
                     continue
 
+                # DeepSeek models have issues with reasoningContent
+                # TODO: Replace with systematic model configuration registry (https://github.com/strands-agents/sdk-python/issues/780)
+                if "deepseek" in self.config["model_id"].lower() and "reasoningContent" in content_block:
+                    dropped_deepseek_reasoning_content = True
+                    continue
+
                 if "toolResult" in content_block:
                     # Create a new content block with only the cleaned toolResult
                     tool_result: ToolResult = content_block["toolResult"]
@@ -327,14 +335,19 @@ def _format_bedrock_messages(self, messages: Messages) -> Messages:
                     # Keep other content blocks as-is
                     cleaned_content.append(content_block)
 
-            # Create new message with cleaned content
-            cleaned_message: Message = Message(content=cleaned_content, role=message["role"])
-            cleaned_messages.append(cleaned_message)
+            # Create new message with cleaned content (skip if empty for DeepSeek)
+            if cleaned_content:
+                cleaned_message: Message = Message(content=cleaned_content, role=message["role"])
+                cleaned_messages.append(cleaned_message)
 
         if filtered_unknown_members:
             logger.warning(
                 "Filtered out SDK_UNKNOWN_MEMBER content blocks from messages, consider upgrading boto3 version"
             )
+        if dropped_deepseek_reasoning_content:
+            logger.debug(
+                "Filtered DeepSeek reasoningContent content blocks from messages - https://api-docs.deepseek.com/guides/reasoning_model#multi-round-conversation"
+            )
 
         return cleaned_messages
 
@@ -386,7 +399,8 @@ def _generate_redaction_events(self) -> list[StreamEvent]:
                 {
                     "redactContent": {
                         "redactAssistantContentMessage": self.config.get(
-                            "guardrail_redact_output_message", "[Assistant output redacted.]"
+                            "guardrail_redact_output_message",
+                            "[Assistant output redacted.]",
                         )
                     }
                 }
@@ -699,7 +713,11 @@ def _find_detected_and_blocked_policy(self, input: Any) -> bool:
 
     @override
     async def structured_output(
-        self, output_model: Type[T], prompt: Messages, system_prompt: Optional[str] = None, **kwargs: Any
+        self,
+        output_model: Type[T],
+        prompt: Messages,
+        system_prompt: Optional[str] = None,
+        **kwargs: Any,
     ) -> AsyncGenerator[dict[str, Union[T, Any]], None]:
         """Get structured output from the model.
 
@@ -714,7 +732,12 @@ async def structured_output(
         """
         tool_spec = convert_pydantic_to_tool_spec(output_model)
 
-        response = self.stream(messages=prompt, tool_specs=[tool_spec], system_prompt=system_prompt, **kwargs)
+        response = self.stream(
+            messages=prompt,
+            tool_specs=[tool_spec],
+            system_prompt=system_prompt,
+            **kwargs,
+        )
         async for event in streaming.process_stream(response):
             yield event
 
diff --git a/tests/strands/models/test_bedrock.py b/tests/strands/models/test_bedrock.py
index 13918b6ea..f2e459bde 100644
--- a/tests/strands/models/test_bedrock.py
+++ b/tests/strands/models/test_bedrock.py
@@ -1392,3 +1392,56 @@ def test_format_request_filters_sdk_unknown_member_content_blocks(model, model_i
 
     for block in content:
         assert "SDK_UNKNOWN_MEMBER" not in block
+
+
+@pytest.mark.asyncio
+async def test_stream_deepseek_filters_reasoning_content(bedrock_client, alist):
+    """Test that DeepSeek models filter reasoningContent from messages during streaming."""
+    model = BedrockModel(model_id="us.deepseek.r1-v1:0")
+
+    messages = [
+        {"role": "user", "content": [{"text": "Hello"}]},
+        {
+            "role": "assistant",
+            "content": [
+                {"text": "Response"},
+                {"reasoningContent": {"reasoningText": {"text": "Thinking..."}}},
+            ],
+        },
+    ]
+
+    bedrock_client.converse_stream.return_value = {"stream": []}
+
+    await alist(model.stream(messages))
+
+    # Verify the request was made with filtered messages (no reasoningContent)
+    call_args = bedrock_client.converse_stream.call_args[1]
+    sent_messages = call_args["messages"]
+
+    assert len(sent_messages) == 2
+    assert sent_messages[0]["content"] == [{"text": "Hello"}]
+    assert sent_messages[1]["content"] == [{"text": "Response"}]
+
+
+@pytest.mark.asyncio
+async def test_stream_deepseek_skips_empty_messages(bedrock_client, alist):
+    """Test that DeepSeek models skip messages that would be empty after filtering reasoningContent."""
+    model = BedrockModel(model_id="us.deepseek.r1-v1:0")
+
+    messages = [
+        {"role": "user", "content": [{"text": "Hello"}]},
+        {"role": "assistant", "content": [{"reasoningContent": {"reasoningText": {"text": "Only reasoning..."}}}]},
+        {"role": "user", "content": [{"text": "Follow up"}]},
+    ]
+
+    bedrock_client.converse_stream.return_value = {"stream": []}
+
+    await alist(model.stream(messages))
+
+    # Verify the request was made with only non-empty messages
+    call_args = bedrock_client.converse_stream.call_args[1]
+    sent_messages = call_args["messages"]
+
+    assert len(sent_messages) == 2
+    assert sent_messages[0]["content"] == [{"text": "Hello"}]
+    assert sent_messages[1]["content"] == [{"text": "Follow up"}]