---
date: "2024-08-22T08:34:51.000Z"
title: "2024-08-22"
tags: ["openrouter","llama3.1"]
draft: false
---

I tried out [OpenRouter](https://openrouter.ai/) for the first time.
My struggles to find an API that hosted `llama3.1-405B` motivated me to try this out.
There are too many companies providing inference APIs to keep track.
OpenRouter seems to be aiming to make all these available from a single place, sort of like [AWS Bedrock](https://aws.amazon.com/bedrock/), but not locked in cloud configuration purgatory.
The first thing I tried was playing a game of Connections with `nousresearch/hermes-3-llama-3.1-405b`.
It didn't get any categories correct for the [2024-08-21](https://www.nytimes.com/svc/connections/v2/2024-08-21.json) puzzle.
OpenRouter's app showcase list is an interesting window into how people are using models.
The dominant themes are

- coding
- characters and roleplay
- writing assistant
- homework

Users of some of these apps generate almost 1B tokens per day according to OpenRouter's stats.
The highest per month (excluding the OpenRouter Playground) is over 30B.

---

Playing with all these models, I've started to crystallize a set of a few things I try to get a sense of the vibes

- play Connections
- ask "hi, who am I speaking with?"
- ask for Python code to read all JSON files from my Desktop and all subfolders
- ask "what is the most recent event you can remember?" which usually gets a response related to the date the training data snapshot was taken