diff --git a/Llama.hs b/Llama.hs
--- a/Llama.hs
+++ b/Llama.hs
@@ -27,7 +27,7 @@
       , "content" .= content m
       ]
 
-data LlamaApplyTemplateRequest = LlamaApplyTemplateRequest
+newtype LlamaApplyTemplateRequest = LlamaApplyTemplateRequest
   { messages :: [LlamaMessage]
   } deriving (Show, Generic)
 instance ToJSON LlamaApplyTemplateRequest
@@ -37,7 +37,7 @@
 --  } deriving (Show)
 
 -- Llama request and response
-data LlamaRequest = LlamaRequest
+newtype LlamaRequest = LlamaRequest
   { prompt :: Text
   } deriving (Show)
 
@@ -49,7 +49,7 @@
   toJSON (LlamaRequest p) =
     object ["prompt" .= p]
 
-data LlamaResponse = LlamaResponse
+newtype LlamaResponse = LlamaResponse
   { generatedText :: Text
   } deriving (Show)
 
@@ -57,9 +57,11 @@
   parseJSON = withObject "LlamaResponse" $ \v -> LlamaResponse
     <$> v .: "content"
 
-applyTemplate :: Manager -> LlamaApplyTemplateRequest -> IO (Maybe Text)
-applyTemplate manager input = do
-  let request = parseRequest_ "http://localhost:8080/apply-template"
+type URL = String
+
+applyTemplate :: URL -> Manager -> LlamaApplyTemplateRequest -> IO (Maybe Text)
+applyTemplate url manager input = do
+  let request = parseRequest_ $ url ++ "/apply-template"
       body = encode input
       req = request { method = "POST"
                     , requestBody = RequestBodyLBS body
@@ -73,9 +75,9 @@
       return Nothing
 
 -- Function to send a message to the Llama model
-sendToLlama :: Manager -> Text -> IO (Maybe Text)
-sendToLlama manager input = do
-  let request = parseRequest_ "http://localhost:8080/completion"
+sendToLlama :: URL -> Manager -> Text -> IO (Maybe Text)
+sendToLlama url manager input = do
+  let request = parseRequest_ $ url ++ "/completion"
       body = encode (LlamaRequest input)
       req = request { method = "POST"
                     , requestBody = RequestBodyLBS body
@@ -88,15 +90,15 @@
       liftIO $ hPutStrLn stderr "Failed to decode Llama response"
       return Nothing
 
-llama :: Text -> IO (Maybe Text)
-llama input = do
+llama :: URL -> Text -> IO (Maybe Text)
+llama url input = do
   manager <- liftIO $ newManager tlsManagerSettings { managerResponseTimeout = responseTimeoutNone }
-  sendToLlama manager input
+  sendToLlama url manager input
 
-llamaTemplated :: LlamaApplyTemplateRequest -> IO (Maybe Text)
-llamaTemplated input = do
+llamaTemplated :: URL -> LlamaApplyTemplateRequest -> IO (Maybe Text)
+llamaTemplated url input = do
   manager <- liftIO $ newManager tlsManagerSettings { managerResponseTimeout = responseTimeoutNone }
-  res <- applyTemplate manager input
+  res <- applyTemplate url manager input
   case res of
-    Just text -> sendToLlama manager text
+    Just text -> sendToLlama url manager text
     _ -> pure Nothing
diff --git a/llama-cpp-haskell.cabal b/llama-cpp-haskell.cabal
--- a/llama-cpp-haskell.cabal
+++ b/llama-cpp-haskell.cabal
@@ -1,6 +1,6 @@
 cabal-version:      2.2
 name:               llama-cpp-haskell
-version:            0.0.1
+version:            0.1
 synopsis:           Haskell bindings for the llama.cpp llama-server
 -- description:
 license:            AGPL-3.0-only
@@ -19,7 +19,7 @@
 Source-repository this
   type:              git
   location:          https://github.com/l29ah/llama-cpp-haskell.git
-  tag:               0.0.1
+  tag:               0.1
 
 common stuff
     ghc-options: -Wall