[Frontend] Delegate to vLLM Omni When `--omni` Passed (#40744)

Signed-off-by: Alex Brooks <albrooks@redhat.com>

vllm/entrypoints/cli/main.py

@@ -7,6 +7,7 @@ to avoid certain eager import breakage."""
 
 import importlib.metadata
 import sys
+from importlib.util import find_spec
 
 from vllm.logger import init_logger
 
@@ -34,6 +35,22 @@ def main():
 
     cli_env_setup()
 
+    # If `--omni` arg is passed to the CLI, delegate to vLLM Omni's entrypoint handling
+    if "--omni" in sys.argv:
+        # NOTE: Check the spec instead of importing directly here, since things could
+        # fail with ImportError due to mismatched versions if things are moved around.
+        spec = find_spec("vllm_omni")
+        if spec is None:
+            logger.error(
+                "--omni flag requires a valid instance of vllm-omni to be installed."
+            )
+            sys.exit(1)
+
+        from vllm_omni.entrypoints.cli.main import main as omni_main
+
+        logger.info("Delegating entrypoint handling to vllm-omni")
+        omni_main()
+    else:
         # For 'vllm bench *': use CPU instead of UnspecifiedPlatform by default
         if len(sys.argv) > 1 and sys.argv[1] == "bench":
             logger.debug(