feat(models): add support for image content blocks in responses

src/strands/event_loop/streaming.py

@@ -248,6 +248,13 @@ def handle_content_block_delta(
             state["redactedContent"] = state.get("redactedContent", b"") + redacted_content
             typed_event = ReasoningRedactedContentStreamEvent(redacted_content=redacted_content, delta=delta_content)
 
+    elif "image" in delta_content:
+        if "image" not in state:
+            state["image"] = []
+
+        state["image"].append(delta_content["image"])
+        typed_event = ModelStreamEvent(delta_content)
+
     return state, typed_event
 
 
@@ -313,6 +320,11 @@ def handle_content_block_stop(state: dict[str, Any]) -> dict[str, Any]:
     elif redacted_content:
         content.append({"reasoningContent": {"redactedContent": redacted_content}})
         state["redactedContent"] = b""
+    elif state.get("image"):
+        # Add all accumulated image blocks to content
+        for image_data in state["image"]:
+            content.append({"image": image_data})
+        state["image"] = []
 
     return state
 
@@ -382,6 +394,7 @@ async def process_stream(
         "current_tool_use": {},
         "reasoningText": "",
         "citationsContent": [],
+        "image": [],
     }
     state["content"] = state["message"]["content"]
 

src/strands/models/bedrock.py

@@ -845,6 +845,13 @@ def _convert_non_streaming_to_streaming(self, response: dict[str, Any]) -> Itera
                         "sourceContent": citation["sourceContent"],
                     }
                     yield {"contentBlockDelta": {"delta": {"citation": citation_metadata}}}
+            elif "image" in content:
+                # Yield image content as a delta
+                yield {
+                    "contentBlockDelta": {
+                        "delta": {"image": content["image"]},
+                    }
+                }
 
             # Yield contentBlockStop event
             yield {"contentBlockStop": {}}

tests/strands/event_loop/test_streaming.py

@@ -1070,3 +1070,42 @@ async def test_stream_messages_normalizes_messages(agenerator, alist):
         {"content": [{"toolUse": {"name": "INVALID_TOOL_NAME"}}], "role": "assistant"},
         {"content": [{"toolUse": {"name": "INVALID_TOOL_NAME"}}], "role": "assistant"},
     ]
+
+
+@pytest.mark.asyncio
+async def test_process_stream_with_image_content():
+    """Test that image content blocks are properly processed in streaming responses."""
+    test_image_data = {
+        "format": "png",
+        "source": {
+            "bytes": "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAADUlEQVR42mNk+M9QDwADhgGAWjR9awAAAABJRU5ErkJggg=="
+        },
+    }
+
+    chunks = [
+        cast(MessageStartEvent, {"messageStart": {"role": "assistant"}}),
+        cast(ContentBlockStartEvent, {"contentBlockStart": {"start": {}}}),
+        cast(ContentBlockDeltaEvent, {"contentBlockDelta": {"delta": {"image": test_image_data}}}),
+        {"contentBlockStop": {}},
+        cast(MessageStopEvent, {"messageStop": {"stopReason": "end_turn"}}),
+        {"metadata": {"usage": {"inputTokens": 10, "outputTokens": 5, "totalTokens": 15}}},
+    ]
+
+    async def async_chunks():
+        for chunk in chunks:
+            yield chunk
+
+    events = []
+    async for event in strands.event_loop.streaming.process_stream(async_chunks()):
+        events.append(event)
+
+    # Verify we got a stop event with the image in the message
+    stop_event = cast(ModelStopReason, events[-1])
+    assert "stop" in stop_event
+
+    stop_reason, message, usage, metrics = stop_event["stop"]
+    assert stop_reason == "end_turn"
+    assert message["role"] == "assistant"
+    assert len(message["content"]) == 1
+    assert "image" in message["content"][0]
+    assert message["content"][0]["image"] == test_image_data