Update app.py

app.py

@@ -1,6 +1,6 @@
 import gradio as gr
 import os
-from huggingface_hub import InferenceClient
+from openai import OpenAI
 from get_machine_from_json import string_to_bsg
 import json
 
@@ -17,17 +17,13 @@ try:
 except:
     pass
 
-# Available models on HF Inference API (confirmed working)
-AVAILABLE_MODELS = [
-    "Qwen/Qwen2.5-72B-Instruct",           # Qwen 72B - Very good performance
-    "meta-llama/Llama-3.2-3B-Instruct",    # Llama 3B - Fast and reliable
-    "mistralai/Mistral-7B-Instruct-v0.3",  # Mistral 7B - Good balance
-    "microsoft/Phi-3-mini-4k-instruct",    # Phi 3 Mini - Small and fast
-]
+# Model configuration - Using HF Router with DeepSeek
+MODEL_NAME = "deepseek-ai/DeepSeek-V3.2-Exp"
+HF_ROUTER_BASE_URL = "https://router.huggingface.co/v1"
 
-# Initialize HF Inference client
+# Initialize OpenAI-compatible client for HF Router
 def create_client():
-    """Create HF Inference Client"""
+    """Create OpenAI-compatible client using HF Router"""
     # Get token from environment variable
     hf_token = os.environ.get("HF_TOKEN")
     
@@ -44,19 +40,10 @@ def create_client():
             "Token should have READ permission and start with 'hf_'"
         )
     
-    # Try models in priority order
-    for model in AVAILABLE_MODELS:
-        try:
-            client = InferenceClient(model=model, token=hf_token)
-            print(f"✅ Successfully connected to model: {model}")
-            return client
-        except Exception as e:
-            print(f"⚠️ Failed to connect to {model}: {e}")
-            continue
-    
-    # Fallback to first model if all failed
-    print(f"⚠️ All models failed, using fallback: {AVAILABLE_MODELS[0]}")
-    return InferenceClient(model=AVAILABLE_MODELS[0], token=hf_token)
+    return OpenAI(
+        base_url=HF_ROUTER_BASE_URL,
+        api_key=hf_token
+    )
 
 def generate_machine(user_prompt, temperature=0.7, max_tokens=4096):
     """
@@ -82,17 +69,20 @@ def generate_machine(user_prompt, temperature=0.7, max_tokens=4096):
             {"role": "user", "content": user_prompt}
         ]
         
-        # Call HF Inference API
+        # Call HF Router API
         response = ""
         try:
-            for message in client.chat_completion(
+            stream = client.chat.completions.create(
+                model=MODEL_NAME,
                 messages=messages,
                 temperature=temperature,
                 max_tokens=max_tokens,
                 stream=True,
-            ):
-                if message.choices[0].delta.content:
-                    response += message.choices[0].delta.content
+            )
+            
+            for chunk in stream:
+                if chunk.choices[0].delta.content:
+                    response += chunk.choices[0].delta.content
                     
         except Exception as api_error:
             error_msg = str(api_error)
@@ -518,7 +508,7 @@ with gr.Blocks(css=custom_css, theme=gr.themes.Soft(), title="🎮 Besiege Machi
     gr.HTML(f"""
         <div class="footer-box">
             <p>
-                🤖 Using <a href="https://huggingface.co/{AVAILABLE_MODELS[0]}" target="_blank">{AVAILABLE_MODELS[0]}</a> (HF Inference API)
+                🤖 Using <a href="https://huggingface.co/{MODEL_NAME}" target="_blank">{MODEL_NAME}</a> (HF Router)
             </p>
             <p>
                 ⚠️ Note: Generated machines may require adjustments in-game