diff --git a/tests/serve/test_sidecar.py b/tests/serve/test_sidecar.py
index f49c1e7af3dc..aa99a3b38215 100644
--- a/tests/serve/test_sidecar.py
+++ b/tests/serve/test_sidecar.py
@@ -8,12 +8,14 @@
 
 import pytest
 
+from dynamo.common.utils.install_media_decoders import VALIDATED_SPECS
 from tests.router.helper import generate_random_suffix
 from tests.serve.common import (
     WORKSPACE_DIR,
     params_with_model_mark,
     run_serve_deployment,
 )
+from tests.serve.conftest import MULTIMODAL_VIDEO_EXPECTED, MULTIMODAL_VIDEO_URL
 from tests.utils.constants import DynamoPortRange
 from tests.utils.engine_metrics import (
     EngineMetrics,
@@ -24,6 +26,7 @@
 from tests.utils.gpu_args import map_cuda_visible_devices
 from tests.utils.payload_builder import (
     chat_logprobs_and_structured_output_payloads,
+    chat_payload,
     disaggregated_kv_routing_payload,
     disaggregated_token_count_payload,
     http_cancellation_payloads,
@@ -219,8 +222,8 @@ def _on_first_gpu(*names: str) -> dict[str, tuple[int, ...]]:
             *http_cancellation_payloads(None, max_tokens=CANCELLATION_MAX_TOKENS),
         ],
     ),
-    "trtllm_aggregated_image": SidecarConfig(
-        name="trtllm_aggregated_image",
+    "trtllm_aggregated_multimodal": SidecarConfig(
+        name="trtllm_aggregated_multimodal",
         directory=trtllm_sidecar_dir,
         script_name="agg.sh",
         marks=[
@@ -231,16 +234,31 @@ def _on_first_gpu(*names: str) -> dict[str, tuple[int, ...]]:
             pytest.mark.requested_trtllm_kv_tokens(32768),
             pytest.mark.timeout(960),
             pytest.mark.post_merge,
+            pytest.mark.installs_extra_dependencies,
         ],
         model="Qwen/Qwen3-VL-2B-Instruct",
         timeout=900,
         delayed_start=60,
-        env={"PYTHONUNBUFFERED": "1"},
+        env={
+            "PYTHONUNBUFFERED": "1",
+            "DYN_MM_ALLOW_INTERNAL": "1",
+            "TRTLLM_MEDIA_ALLOW_PRIVATE_URLS": "1",
+            "DYN_TEST_ONLY_PIP_INSTALL": VALIDATED_SPECS["opencv-python-headless"],
+        },
         request_payloads=[
             multimodal_payload_default(
                 text="Describe what you see in this image.",
                 expected_response=["mountain", "rock", "trees", "road"],
-            )
+            ),
+            chat_payload(
+                [
+                    {"type": "text", "text": "Describe the video in detail."},
+                    {"type": "video_url", "video_url": {"url": MULTIMODAL_VIDEO_URL}},
+                ],
+                expected_response=MULTIMODAL_VIDEO_EXPECTED,
+                max_tokens=100,
+                temperature=0.0,
+            ),
         ],
     ),
     "trtllm_disaggregated": SidecarConfig(
@@ -513,6 +531,8 @@ def sidecar_config_test(request, monkeypatch, discovery_backend):
     config = sidecar_configs[request.param]
     if config.name.startswith("trtllm_"):
         require_trtllm_openengine()
+    if config.name == "trtllm_aggregated_multimodal":
+        request.getfixturevalue("image_server")
     monkeypatch.setenv("MODEL", config.model)
     monkeypatch.setenv("DYN_DISCOVERY_BACKEND", discovery_backend)
     monkeypatch.setenv("DYN_NAMESPACE", f"sidecar-{generate_random_suffix()}")