diff --git a/llama_cpp/server/__main__.py b/llama_cpp/server/__main__.py
index a294ebf8a..df54e0326 100644
--- a/llama_cpp/server/__main__.py
+++ b/llama_cpp/server/__main__.py
@@ -96,5 +96,8 @@ def parse_bool_arg(arg):
     app = create_app(settings=settings)
 
     uvicorn.run(
-        app, host=os.getenv("HOST", settings.host), port=int(os.getenv("PORT", settings.port))
+        app,
+        host=os.getenv("HOST", settings.host),
+        port=int(os.getenv("PORT", settings.port)),
+        limit_concurrency=int(os.getenv("LIMIT_CONCURRENCY", settings.limit_concurrency))
     )
diff --git a/llama_cpp/server/app.py b/llama_cpp/server/app.py
index bec956147..031f0560a 100644
--- a/llama_cpp/server/app.py
+++ b/llama_cpp/server/app.py
@@ -157,6 +157,10 @@ class Settings(BaseSettings):
     )
     # Server Params
     host: str = Field(default="localhost", description="Listen address")
+    limit_concurrency: int = Field(
+        default=2,
+        description="Number of concurrent processes + 1"
+    )
     port: int = Field(default=8000, description="Listen port")
     interrupt_requests: bool = Field(
         default=True,