[Feature] Support text encoder Disaggregation for H3 (#5885)

Signed-off-by: Canlin Guo <canlinguosdu@gmail.com>
Signed-off-by: lishunyang12 <lishunyang12@163.com>
Signed-off-by: yuanwu <yuan.wu@intel.com>
Co-authored-by: lishunyang12 <lishunyang12@163.com>
Co-authored-by: yuanwu <yuan.wu@intel.com>

tests/entrypoints/openai_api/test_video_server.py

@@ -64,7 +64,13 @@ class MockVideoResult:
 
 class FakeAsyncOmni:
     def __init__(self):
-        self.stage_configs = [SimpleNamespace(stage_type="diffusion")]
+        self.stage_configs = [
+            SimpleNamespace(
+                stage_type="diffusion",
+                final_output=True,
+                final_output_type="video",
+            )
+        ]
         self.default_sampling_params_list = [OmniDiffusionSamplingParams()]
         self.model_class_name = "WanPipeline"
         self.captured_prompt = None
@@ -828,6 +834,18 @@ def test_multi_video_generation_preserves_uploaded_files_until_generation(
     assert engine.captured_sampling_params_list[0].extra_args["duration"] == 15.0
 
 
+def test_mixed_reference_capability_uses_model_metadata_when_config_defaults_false(test_client):
+    handler = test_client.app.state.openai_serving_video
+    handler._engine_client.model_class_name = None
+    handler._engine_client.stage_configs = [
+        SimpleNamespace(engine_args={"model_class_name": "MiniMaxH3TextEncoder"}),
+        SimpleNamespace(engine_args={"model_class_name": "MiniMaxH3Pipeline"}),
+    ]
+    handler._stage_configs = handler._engine_client.stage_configs
+
+    assert handler.supports_mixed_reference_inputs
+
+
 def test_decode_video_bytes_can_keep_first_frames():
     from vllm_omni.entrypoints.openai.video_api_utils import _decode_video_bytes