diff --git a/python/sglang/srt/managers/tokenizer_manager.py b/python/sglang/srt/managers/tokenizer_manager.py
index 199ea7c3a..3f25ad560 100644
--- a/python/sglang/srt/managers/tokenizer_manager.py
+++ b/python/sglang/srt/managers/tokenizer_manager.py
@@ -428,7 +428,7 @@ class TokenizerManager:
 
     async def _get_pixel_values(self, image_data):
         if isinstance(image_data, list) and len(image_data) > 0:
-            return await self._get_pixel_values_internal(image_data[0])
+            return await self._get_pixel_values_internal(image_data)
         elif isinstance(image_data, str):
             return await self._get_pixel_values_internal(image_data)
         else:
diff --git a/test/srt/test_vision_openai_server.py b/test/srt/test_vision_openai_server.py
index a34571776..0003e4776 100644
--- a/test/srt/test_vision_openai_server.py
+++ b/test/srt/test_vision_openai_server.py
@@ -114,8 +114,8 @@ class TestOpenAIVisionServer(unittest.TestCase):
         text = response.choices[0].message.content
         assert isinstance(text, str)
         print(text)
-        assert "man" in text or "cab" in text, text
-        # assert "logo" in text, text
+        assert "man" in text and "taxi" in text, text
+        assert "logo" in text, text
         assert response.id
         assert response.created
         assert response.usage.prompt_tokens > 0