Each mark is one effort level of one model, measured on the same task set by Artificial Analysis. Up means more capable, right means more expensive per task, so a line that sits higher at the same cost gets more for the money. This page rebuilds from live data each time you run /llm-chart.
Hover or tab to a mark for its numbers. A ring marks a setting whose score or cost changed since the last run. A 1-point gap on this index is small.
| Model | Effort | Index | Cost / task | Output tokens (index run) | First answer token |
|---|