Dominik Weckmüller

Looks like @openrouter is the first to offer a properly-hosted Reflection Llama 3.1 70B. It's time to fucking build. Here's the code to use: ``` import requests import json response = requests.post( url="openrouter.ai/api/v1/chat/co…", headers={ "Authorization": f"Bearer YOUR_API_KEY", }, data=json.dumps({ "model": "mattshumer/reflection-70b", "messages": [ { "role": "system", "content": "You are a world-class AI system, capable of complex reasoning and reflection. Reason through the query inside tags, and then provide your final response inside

tags. If you detect that you made a mistake in your reasoning at any point, correct yourself inside tags." }, { "role": "user", "content": "How many Rs are in strawberry?" } ] }) ) ``` Note that this is FP8, so it'll be a good bit less accurate than the full fat version, but should be somewhat close.