diff --git a/g4f/Provider/needs_auth/hf/HuggingFaceAPI.py b/g4f/Provider/needs_auth/hf/HuggingFaceAPI.py
index 538b6a0c..2ce1609d 100644
--- a/g4f/Provider/needs_auth/hf/HuggingFaceAPI.py
+++ b/g4f/Provider/needs_auth/hf/HuggingFaceAPI.py
@@ -89,7 +89,16 @@ class HuggingFaceAPI(OpenaiTemplate):
             raise ModelNotFoundError(f"Model is not supported: {model} in: {cls.__name__}")
         error = None
         for provider_key in provider_mapping:
-            api_path = "groq/openai/v1" if provider_key == "groq" else "novita/v3/openai" if provider_key == "novita" else f"{provider_key}/v1"
+            if provider_key == "zai-org":
+                api_path = "zai-org/api/paas/v4"
+            elif provider_key == "novita":
+                api_path = "novita/v3/openai"
+            elif provider_key == "groq":
+                api_path = "groq/openai/v1"
+            elif provider_key == "hf-inference":
+                api_path = f"{provider_key}/models/{model}/v1"
+            else:
+                api_path = f"{provider_key}/v1"
             api_base = f"https://router.huggingface.co/{api_path}"
             task = provider_mapping[provider_key]["task"]
             if task != "conversational":
diff --git a/g4f/client/__init__.py b/g4f/client/__init__.py
index 0c6dcd78..ac8a42bb 100644
--- a/g4f/client/__init__.py
+++ b/g4f/client/__init__.py
@@ -308,7 +308,6 @@ class Completions:
         response_format: Optional[dict] = None,
         max_tokens: Optional[int] = None,
         stop: Optional[Union[list[str], str]] = None,
-        api_key: Optional[str] = None,
         ignore_stream: Optional[bool] = False,
         raw: Optional[bool] = False,
         **kwargs
@@ -337,7 +336,8 @@ class Completions:
                 proxy=self.client.proxy if proxy is None else proxy,
                 max_tokens=max_tokens,
                 stop=stop,
-                api_key=self.client.api_key if api_key is None else api_key
+                api_key=self.client.api_key,
+                api_base=self.client.base_url
             ),
             **kwargs
         )
@@ -646,7 +646,6 @@ class AsyncCompletions:
         response_format: Optional[dict] = None,
         max_tokens: Optional[int] = None,
         stop: Optional[Union[list[str], str]] = None,
-        api_key: Optional[str] = None,
         ignore_stream: Optional[bool] = False,
         raw: Optional[bool] = False,
         **kwargs
@@ -675,7 +674,8 @@ class AsyncCompletions:
                 proxy=self.client.proxy if proxy is None else proxy,
                 max_tokens=max_tokens,
                 stop=stop,
-                api_key=self.client.api_key if api_key is None else api_key
+                api_key=self.client.api_key,
+                api_base=self.client.base_url
             ),
             **kwargs
         )
diff --git a/g4f/client/types.py b/g4f/client/types.py
index 785f3d8b..eb2400ac 100644
--- a/g4f/client/types.py
+++ b/g4f/client/types.py
@@ -15,11 +15,13 @@ class Client():
         self,
         api_key: str = None,
         proxies: Proxies = None,
+        base_url: str = None,
         **kwargs
     ) -> None:
         self.api_key: str = api_key
         self.proxies = proxies 
         self.proxy: str = self.get_proxy()
+        self.base_url: str = base_url
 
     def get_proxy(self) -> Union[str, None]:
         if isinstance(self.proxies, str):