[Model][VLM] Add Kimi-VL model support (#16387)

Signed-off-by: courage17340 <courage17340@163.com>

vllm/entrypoints/chat_utils.py

@@ -512,6 +512,8 @@ class BaseMultiModalItemTracker(ABC, Generic[_T]):
                 return "<|fim_prefix|><|img|><|fim_suffix|>"
             if model_type == "gemma3":
                 return "<start_of_image>"
+            if model_type == "kimi_vl":
+                return "<|media_start|>image<|media_content|><|media_pad|><|media_end|>" # noqa: E501
 
             raise TypeError(f"Unknown {modality} model type: {model_type}")
         elif modality == "audio":