app.py

import os
import gradio as gr
import requests

# Get the Hugging Face API key from Spaces secrets.
HF_API_KEY = os.getenv("HF_API_KEY")

# Model endpoints on Hugging Face
MODEL_ENDPOINTS = {
    "Qwen2.5-72B-Instruct": "https://api-inference.huggingface.co/models/Qwen/Qwen2.5-72B-Instruct",
    "Llama3.3-70B-Instruct": "https://api-inference.huggingface.co/models/meta-llama/Llama-3.3-70B-Instruct",
    "Qwen2.5-Coder-32B-Instruct": "https://api-inference.huggingface.co/models/Qwen/Qwen2.5-Coder-32B-Instruct",
}

def query_model(prompt, model_endpoint):
    """
    Query a model via Hugging Face Inference API using a requests.post call.
    This assumes an OpenAI-compatible endpoint structure.
    """
    headers = {
        "Authorization": f"Bearer {HF_API_KEY}",
        "Content-Type": "application/json"
    }
    data = {
        "messages": [{"role": "user", "content": prompt}],
        "max_tokens": 512,
        "temperature": 0.7
    }
    
    response = requests.post(model_endpoint, headers=headers, json=data)
    try:
        result = response.json()
    except Exception:
        return f"Error: Unable to parse JSON. Response: {response.text}"
    
    if "error" in result:
        return f"Error: {result['error']}"
    
    try:
        return result["choices"][0]["message"]["content"]
    except Exception:
        return f"Error: Unexpected response format: {result}"

def chat_with_models(user_input, history):
    responses = []
    for model_name, endpoint in MODEL_ENDPOINTS.items():
        model_response = query_model(user_input, endpoint)
        responses.append(f"**{model_name}**: {model_response}")
    combined_answer = "\n\n".join(responses)
    history.append((user_input, combined_answer))
    return history, history

with gr.Blocks() as demo:
    gr.Markdown("# Multi-LLM Chatbot using Hugging Face Inference API")
    chatbot = gr.Chatbot()
    msg = gr.Textbox(label="Your Message")
    clear = gr.Button("Clear")

    def clear_chat():
        return [], []

    msg.submit(fn=chat_with_models, inputs=[msg, chatbot], outputs=[chatbot, chatbot])
    clear.click(fn=clear_chat, outputs=[chatbot, chatbot])

demo.launch()