diff --git a/gradio/app.py b/gradio/app.py
index 0cd1b37d81..40a0a56552 100644
--- a/gradio/app.py
+++ b/gradio/app.py
@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 from fastapi import FastAPI, Request, Depends, HTTPException, status
+from fastapi.concurrency import run_in_threadpool
 from fastapi.middleware.cors import CORSMiddleware
 from fastapi.responses import JSONResponse, HTMLResponse, FileResponse
 from fastapi.security import OAuth2PasswordRequestForm
@@ -37,6 +38,7 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
+
 templates = Jinja2Templates(directory=STATIC_TEMPLATE_LIB)
 
 
@@ -178,23 +180,28 @@ async def predict(
     if body.get("example_id") != None:
         example_id = body["example_id"]
         if app.interface.cache_examples:
-            prediction = load_from_cache(app.interface, example_id)
+            prediction = await run_in_threadpool(
+                load_from_cache, app.interface, example_id)
             durations = None
         else:
-            prediction, durations = process_example(app.interface, example_id)
+            prediction, durations = await run_in_threadpool(
+                process_example, app.interface, example_id)
     else:
         raw_input = body["data"]
         if app.interface.show_error:
             try:
-                prediction, durations = app.interface.process(raw_input)
+                prediction, durations = await run_in_threadpool(
+                    app.interface.process, raw_input)
             except BaseException as error:
                 traceback.print_exc()
                 return JSONResponse(content={"error": str(error)}, 
                                     status_code=500)
         else:
-            prediction, durations = app.interface.process(raw_input)
+            prediction, durations = await run_in_threadpool(
+                app.interface.process, raw_input)
         if app.interface.allow_flagging == "auto":
-            flag_index = app.interface.flagging_callback.flag(
+            flag_index = await run_in_threadpool(
+                app.interface.flagging_callback.flag,
                 app.interface, raw_input, prediction,
                 flag_option="" if app.interface.flagging_options else None, 
                 username=username)
@@ -216,7 +223,8 @@ async def flag(
         await utils.log_feature_analytics(app.interface.ip_address, 'flag')
     body = await request.json()
     data = body['data']
-    app.interface.flagging_callback.flag(
+    await run_in_threadpool(
+        app.interface.flagging_callback.flag,
         app.interface, data['input_data'], data['output_data'], 
         flag_option=data.get("flag_option"), flag_index=data.get("flag_index"),
         username=username)
@@ -229,8 +237,8 @@ async def interpret(request: Request):
         await utils.log_feature_analytics(app.interface.ip_address, 'interpret')
     body = await request.json()
     raw_input = body["data"]
-    interpretation_scores, alternative_outputs = app.interface.interpret(
-        raw_input)
+    interpretation_scores, alternative_outputs = await run_in_threadpool(
+        app.interface.interpret, raw_input)
     return {
         "interpretation_scores": interpretation_scores,
         "alternative_outputs": alternative_outputs
diff --git a/gradio/networking.py b/gradio/networking.py
index 932ea7beb2..33b1a512c4 100644
--- a/gradio/networking.py
+++ b/gradio/networking.py
@@ -147,7 +147,7 @@ def start_server(
         app.queue_thread.start()
     if interface.save_to is not None:  # Used for selenium tests
         interface.save_to["port"] = port
-                
+    
     config = uvicorn.Config(app=app, port=port, host=server_name, 
                             log_level="warning")
     server = Server(config=config)