diff --git a/frigate/detectors/detection_runners.py b/frigate/detectors/detection_runners.py
index 9ad151398..9a9c1865b 100644
--- a/frigate/detectors/detection_runners.py
+++ b/frigate/detectors/detection_runners.py
@@ -3,6 +3,7 @@
 import json
 import logging
 import os
+import threading
 from abc import ABC, abstractmethod
 from typing import Any
 
@@ -330,9 +331,28 @@ class ZmqIpcRunner(BaseModelRunner):
         self._socket.setsockopt(zmq.LINGER, linger_ms)
         self._socket.connect(self._endpoint)
         self._model_ready = False
+        self._io_lock = threading.Lock()
+
+    @staticmethod
+    def is_complex_model(model_type: str) -> bool:
+        # Import here to avoid circular imports
+        from frigate.detectors.detector_config import ModelTypeEnum
+        from frigate.embeddings.types import EnrichmentModelTypeEnum
+
+        return model_type in [
+            ModelTypeEnum.yolonas.value,
+            EnrichmentModelTypeEnum.paddleocr.value,
+            EnrichmentModelTypeEnum.jina_v1.value,
+            EnrichmentModelTypeEnum.jina_v2.value,
+        ]
 
     def get_input_names(self) -> list[str]:
-        return ["input"]
+        if "vision" in self.model_name:
+            return ["pixel_values"]
+        elif "arcface" in self.model_name:
+            return ["data"]
+        else:
+            return ["input"]
 
     def get_input_width(self) -> int:
         # Not known/required for ZMQ forwarding
@@ -373,8 +393,9 @@ class ZmqIpcRunner(BaseModelRunner):
         header = self._build_header(tensor)
         payload = memoryview(tensor.tobytes(order="C"))
         try:
-            self._socket.send_multipart([header, payload])
-            frames = self._socket.recv_multipart()
+            with self._io_lock:
+                self._socket.send_multipart([header, payload])
+                frames = self._socket.recv_multipart()
         except zmq.Again as e:
             raise TimeoutError("ZMQ detector request timed out") from e
         except zmq.ZMQError as e:
@@ -392,7 +413,8 @@ class ZmqIpcRunner(BaseModelRunner):
         """
         # Check model availability
         req = {"model_request": True, "model_name": self.model_name}
-        self._socket.send_multipart([json.dumps(req).encode("utf-8")])
+        with self._io_lock:
+            self._socket.send_multipart([json.dumps(req).encode("utf-8")])
 
         # Temporarily extend timeout for model ops
         original_rcv = self._socket.getsockopt(zmq.RCVTIMEO)
@@ -425,7 +447,10 @@ class ZmqIpcRunner(BaseModelRunner):
             return False
 
         header = {"model_data": True, "model_name": self.model_name}
-        self._socket.send_multipart([json.dumps(header).encode("utf-8"), model_bytes])
+        with self._io_lock:
+            self._socket.send_multipart(
+                [json.dumps(header).encode("utf-8"), model_bytes]
+            )
 
         original_rcv2 = self._socket.getsockopt(zmq.RCVTIMEO)
         try:
@@ -575,6 +600,10 @@ def get_optimized_runner(
         if rknn_path:
             return RKNNModelRunner(rknn_path)
 
+    if device == "ZMQ" and not ZmqIpcRunner.is_complex_model(model_type):
+        logger.info(f"Using ZMQ detector model {model_path}")
+        return ZmqIpcRunner(model_path, model_type, **kwargs)
+
     providers, options = get_ort_providers(device == "CPU", device, **kwargs)
 
     if providers[0] == "CPUExecutionProvider":