Create app.py

app.py

@@ -0,0 +1,52 @@
+import os
+import gradio as gr
+from huggingface_hub import InferenceClient
+
+# Initialize the client pointing to MoonshotAI Kimi-K3
+# Automatically grabs the secret token from the Space environment
+client = InferenceClient(
+    model="moonshotai/Kimi-K3",
+    token=os.environ.get("HF_TOKEN")
+)
+
+def respond(message, chat_history):
+    # Format system and chat history for the agentic model
+    messages = [{"role": "system", "content": "You are Kimi-K3, a native multimodal agentic frontier model developed by Moonshot AI."}]
+    
+    for val in chat_history:
+        if val[0]:
+            messages.append({"role": "user", "content": val[0]})
+        if val[1]:
+            messages.append({"role": "assistant", "content": val[1]})
+            
+    messages.append({"role": "user", "content": message})
+    
+    # Stream the output chunks safely
+    response = ""
+    try:
+        for message_chunk in client.chat_completion(
+            messages,
+            max_tokens=2048,
+            stream=True,
+            temperature=0.7,
+            top_p=0.95,
+        ):
+            token = message_chunk.choices[0].delta.content
+            if token:
+                response += token
+                yield response
+    except Exception as e:
+        yield f"Error calling Hugging Face Inference Provider: {str(e)}. Ensure your HF_TOKEN is configured correctly in Space Secrets."
+
+# Build a clean, modern Gradio Chat interface
+demo = gr.ChatInterface(
+    fn=respond,
+    title="🤖 Kimi-K3 Demo Space",
+    description="Interface for Moonshot AI's Kimi-K3 2.8T Parameter Frontier Model.",
+    examples=["Write an optimized GPU kernel processing loop.", "Draft a comprehensive research document comparing MoE architectures."],
+    textbox=gr.Textbox(placeholder="Ask Kimi-K3 anything...", container=False, scale=7),
+)
+
+if __name__ == "__main__":
+    demo.launch()
+