Kuzco Inference - Interactive Terminal (mixtral, non-streaming, python)

kuzco.py

import asyncio
import aiohttp
import json
import sys

API_KEY = '[INSERT YOUR KUZCO API KEY HERE]'
BASE_URL = 'https://relay.kuzco.xyz/v1'

async def main():
    messages = []
    print("\n")
    while True:
        user_input = input("User: ")
        if user_input.lower() == 'exit':
            break
        messages.append({'role': 'user', 'content': user_input + '\n'})
        
        timeout = aiohttp.ClientTimeout(total=5*60)
        async with aiohttp.ClientSession(timeout=timeout) as session:
            try:
                async with session.post(
                    f'{BASE_URL}/chat/completions',
                    headers={
                        'Authorization': f'Bearer {API_KEY}',
                        'Content-Type': 'application/json'
                    },
                    json={
                        'messages': messages,
                        'model': 'mistral',
                        'stream': False
                    }
                ) as response:
                    content = await response.text()
                    print(f"\nKuzco (Mistral):\n\n{json.loads(content).get('choices', [{}])[0].get('message', {}).get('content', '').lstrip()}\n\n")
            except asyncio.TimeoutError:
                print("Request timed out")
            except Exception as e:
                print(f"An error occurred: {e}")

def run():
    asyncio.run(main())

if __name__ == '__main__':
    run()