Over the holidays I worked on a few small AI projects and got frustrated with the hassle of choosing a model that's good enough for the job without massively overpaying.

So I built a visualization for the models’ performance vs price:

- Performance comes from LM Arena's ELO, by category (overall, coding, etc.).

- Pricing comes from OpenRouter (USD per 1M tokens).

- The chart highlights the Pareto frontier, so you can see which models are optimal at each price point.

Feedback welcome