Show HN: Models Pie – LLMs ranked by fast / cheap / good tradeoffs (modelspie.com)

4 points by willks ↗ HN
Hey HN, I couldn’t find a good way to rank the current LLM models based on the priorities I have for different tasks, so I built my own.

You set your own trade-off priorities between fast, cheap, and good by dragging two handles around a pie chart to weight the three priorities, and a ranked table re-sorts live underneath. The data comes from BenchLM, OpenRouter and Artificial Analysis. Hope it's useful.

1 comment

[ 4.2 ms ] story [ 15.3 ms ] thread
Super cool, I'm always interested in super high speed + quality. So I picked 50-50.

I'm trying cursor/composer mostly because it seems way faster than azure/aws/gcp

Anyways I'm probably doing, because most peers tell me the pick for intelligence and then let it run unattended "for ever" so ...