AA Long Context Reasoning vs Completion Time

Each model's AA Long Context Reasoning score against its estimated completion time (time to last token) for the selected input and output lengths. Timing is the site's per-model estimate, not a published benchmark run time. Color = provider.

AA Long Context Reasoning vs Completion Time
Controls:
Input10K tok
Output1K tok
Score1:1Completion Time

This chart is part of the AA Long Context Reasoning benchmark page, which adds the model table, sources and how to read each view.