Model of choice

Some data about LLM usage and revenues

• the one with the most tokens

• Claude and Cursor ftw

• Anthropic’s big revenue goals

---

👉👉👉Reminder to sign up for the Weekly Recap only, if daily emails is too much. Find me on twitter, for more fun. 

---

Model of choice

A Friday fun-day look at the top LLM models (by some measures).

The one with the most tokens

I’m not sure how good this data is, but it’s pretty fun.

If you want to know which LLM models are most popular, one way to do it is count tokens—the base unit of text that LLMs process and output.

By token count, Anthropic’s Claude is running away from the field:

OpenRouter

Claude 3.5 Sonnet and Sonnet Unsupervised took the top-spots, with Gemini models nipping at their heels.

I’m not really sure what to make of this, except that perhaps it confirms some of the scuttlebutt that Claude is the best LLM for coding, and coding is the best use-case for AI, and Cursor (which uses Claude, among others) is the most popular app for AI coding.

Claude and Cursor FTW

The scuttlebutt is born out by the transaction data, as well.