Fix model

app.py

@@ -70,50 +70,17 @@ class OmniAPIClient:
         return content_parts
     
     def get_available_models(self, api_key: str = "") -> Tuple[bool, List[str]]:
-        """Fetch available models from the API"""
-        try:
-            # print(f"DEBUG: Fetching models from: {self.models_endpoint}")  # Debug line (commented out)
-            headers = {"Content-Type": "application/json"}
-            if api_key:
-                headers["Authorization"] = f"Bearer {api_key}"
-            
-            response = requests.get(
-                self.models_endpoint,
-                headers=headers,
-                timeout=10
-            )
-            
-            if response.status_code == 200:
-                try:
-                    data = response.json()
-                    # Handle different response formats
-                    if "data" in data and isinstance(data["data"], list):
-                        # OpenAI-style format: {"data": [{"id": "model1"}, {"id": "model2"}]}
-                        models = [model.get("id", "") for model in data["data"] if model.get("id")]
-                    elif "models" in data and isinstance(data["models"], list):
-                        # Custom format: {"models": ["model1", "model2"]}
-                        models = data["models"]
-                    elif isinstance(data, list):
-                        # Direct list format: ["model1", "model2"]
-                        models = data
-                    else:
-                        # Fallback: try to extract any string values
-                        models = []
-                        if isinstance(data, dict):
-                            for key, value in data.items():
-                                if isinstance(value, list):
-                                    models.extend([str(item) for item in value if item])
-                    
-                    return True, models if models else ["qwen/qwen3-235b-a22b-instruct-2507"]  # fallback model
-                except json.JSONDecodeError:
-                    return False, ["qwen/qwen3-235b-a22b-instruct-2507"]
-            else:
-                return False, ["qwen/qwen3-235b-a22b-instruct-2507"]
-                
-        except (requests.exceptions.Timeout, requests.exceptions.ConnectionError):
-            return False, ["qwen/qwen3-235b-a22b-instruct-2507"]
-        except Exception:
-            return False, ["qwen/qwen3-235b-a22b-instruct-2507"]
+        """Return fixed set of available models"""
+        # Fixed set of models as requested
+        fixed_models = [
+            "typhoon-ocr-preview",
+            "openai/gpt-5",
+            "meta-llama/llama-4-maverick",
+            "qwen/qwen3-235b-a22b-instruct-2507",
+            "gemini/gemini-2.5-pro",
+            "gemini/gemini-2.5-flash"
+        ]
+        return True, fixed_models
 
     def send_chat_completion(self, text: str, files: List[str], api_key: str = "", model: str = "qwen/qwen3-235b-a22b-instruct-2507", max_tokens: int = 16384, stream: bool = False) -> Tuple[bool, Any]:
         """Send chat completion request to the API"""
@@ -180,20 +147,17 @@ def create_ui():
     """Create the Gradio UI"""
     
     def fetch_models(base_url, api_key):
-        """Fetch available models from the API"""
-        if not base_url:
-            return gr.Dropdown(choices=["qwen/qwen3-235b-a22b-instruct-2507"], value="qwen/qwen3-235b-a22b-instruct-2507")
-        
-        try:
-            client = OmniAPIClient(base_url)
-            success, models = client.get_available_models(api_key)
-            
-            if success and models:
-                return gr.Dropdown(choices=models, value=models[0] if models else "qwen/qwen3-235b-a22b-instruct-2507")
-            else:
-                return gr.Dropdown(choices=["qwen/qwen3-235b-a22b-instruct-2507"], value="qwen/qwen3-235b-a22b-instruct-2507")
-        except Exception:
-            return gr.Dropdown(choices=["qwen/qwen3-235b-a22b-instruct-2507"], value="qwen/qwen3-235b-a22b-instruct-2507")
+        """Return fixed set of available models"""
+        # Fixed set of models as requested
+        fixed_models = [
+            "typhoon-ocr-preview",
+            "openai/gpt-5",
+            "meta-llama/llama-4-maverick",
+            "qwen/qwen3-235b-a22b-instruct-2507",
+            "gemini/gemini-2.5-pro",
+            "gemini/gemini-2.5-flash"
+        ]
+        return gr.Dropdown(choices=fixed_models, value="qwen/qwen3-235b-a22b-instruct-2507")
     
     def send_request(base_url, api_key, model, max_tokens, text, files):
         """Handle request submission"""
@@ -238,8 +202,12 @@ def create_ui():
                         # Check if model contains 'typhoon'
                         if "typhoon" in model.lower():
                             try:
-                                # Try to get natural_text first
-                                assistant_reply = choice["message"]["content"]["natural_text"]
+                                # If the text is { "natural_text": "..." }, extract the natural_text as json object
+                                json_content = json.loads(choice["message"]["content"])
+                                if "natural_text" in json_content:
+                                    assistant_reply = json_content["natural_text"]
+                                else:
+                                    assistant_reply = choice["message"]["content"]
                             except (KeyError, TypeError):
                                 # Fallback to content if natural_text is not available
                                 assistant_reply = choice["message"]["content"]
@@ -308,12 +276,17 @@ def create_ui():
                with gr.Column(scale=3):
                    model = gr.Dropdown(
                        label="Model",
-                       choices=["qwen/qwen3-235b-a22b-instruct-2507"],
+                       choices=[
+                           "typhoon-ocr-preview",
+                           "openai/gpt-5",
+                           "meta-llama/llama-4-maverick",
+                           "qwen/qwen3-235b-a22b-instruct-2507",
+                           "gemini/gemini-2.5-pro",
+                           "gemini/gemini-2.5-flash"
+                       ],
                        value="qwen/qwen3-235b-a22b-instruct-2507",
                        interactive=True
                    )
-               with gr.Column(scale=1):
-                   refresh_models_btn = gr.Button("🔄", size="sm")
                with gr.Column(scale=2):
                    max_tokens = gr.Number(
                        label="Max Tokens",
@@ -408,40 +381,20 @@ def create_ui():
            outputs=[text_input, status_output, response_output, file_upload]
        )
         
-       # Refresh models when button is clicked
-       refresh_models_btn.click(
+        
+       # Allow Enter key to submit (when text input is focused)
+       text_input.submit(
+           fn=send_request,
+           inputs=[base_url, api_key, model, max_tokens, text_input, file_upload],
+           outputs=[status_output, response_output]
+       )
+        
+       # Preload models when interface loads
+       interface.load(
            fn=fetch_models,
            inputs=[base_url, api_key],
            outputs=[model]
        )
-        
-        # Auto-refresh models when base URL changes
-        base_url.blur(
-            fn=fetch_models,
-            inputs=[base_url, api_key],
-            outputs=[model]
-        )
-        
-        # Auto-refresh models when API key changes
-        api_key.blur(
-            fn=fetch_models,
-            inputs=[base_url, api_key],
-            outputs=[model]
-        )
-        
-        # Allow Enter key to submit (when text input is focused)
-        text_input.submit(
-            fn=send_request,
-            inputs=[base_url, api_key, model, max_tokens, text_input, file_upload],
-            outputs=[status_output, response_output]
-        )
-        
-        # Preload models when interface loads
-        interface.load(
-            fn=fetch_models,
-            inputs=[base_url, api_key],
-            outputs=[model]
-        )
     
     return interface
 
@@ -453,10 +406,10 @@ if __name__ == "__main__":
     # Launch with custom settings
     demo.launch(
         server_name="127.0.0.1",  # Use localhost instead of 0.0.0.0
-        server_port=7890,         # Use different port to avoid conflicts
+        server_port=7892,         # Use different port to avoid conflicts
         share=False,              # Set to True to create public link
         debug=True,              # Disable debug mode to reduce console errors
         show_error=True,          # Show detailed error messages
-        inbrowser=True,           # Auto-open in browser
+        inbrowser=False,           # Auto-open in browser
         prevent_thread_lock=False # Ensure proper threading
     )