Update

devplacepy/services/deepsearch/llm.py

@@ -5,6 +5,8 @@ from __future__ import annotations
 import logging
 import time
 
+import httpx
+
 from devplacepy import stealth
 from devplacepy.config import INTERNAL_GATEWAY_URL, INTERNAL_MODEL
 
@@ -13,6 +15,15 @@ logger = logging.getLogger(__name__)
 CHAT_TIMEOUT_SECONDS = 120.0
 DEFAULT_MAX_TOKENS = 1200
 
+_gateway_client: httpx.AsyncClient | None = None
+
+
+def gateway_client() -> httpx.AsyncClient:
+    global _gateway_client
+    if _gateway_client is None or _gateway_client.is_closed:
+        _gateway_client = stealth.stealth_async_client(timeout=CHAT_TIMEOUT_SECONDS)
+    return _gateway_client
+
 
 async def request_completion(
     messages: list[dict],
@@ -33,10 +44,12 @@ async def request_completion(
     headers = {
         "Authorization": f"Bearer {api_key}",
         "Content-Type": "application/json",
+        "X-App-Reference": "devplace-deepsearch-v-1-0-0",
     }
     start: float = time.monotonic()
-    async with stealth.stealth_async_client(timeout=timeout) as client:
-        response = await client.post(gateway_url, json=payload, headers=headers)
+    response = await gateway_client().post(
+        gateway_url, json=payload, headers=headers, timeout=timeout
+    )
     elapsed_ms: int = int((time.monotonic() - start) * 1000)
     if response.status_code >= 400:
         raise RuntimeError(f"chat gateway returned {response.status_code}")