Skip to content
Trending storyDeveloping

Arena adds per-token and per-minute efficiency views to Agent Arena

1 article1 sourcesince Nov 2Last article 2h ago ·

Overview

AISummary of 1 article

Arena says Agent Arena now ranks agents by performance per output token and per minute of wait, in addition to task spend.

In a post announcing the change, Arena says the same level of intelligence can cost 10 times more, need six times more output tokens, and take twice as long. Its usage plots show cost rises most steeply with intelligence, while output tokens barely predict the bill.

Written by AI from the articles below · updated Nov 2, 11:17 AM ET

Check the sources:

Article timeline

The articles in this story. Times are ET.

Nov 2
  1. Arena.aiOfficial
    Arena adds token and latency efficiency views to Agent Arena

    AIArena's Agent Arena now shows which agents deliver the best performance per output token and per minute of waiting. The post says the same intelligence can cost 10x more, need 6x more output tokens, and take twice as long. Its usage plots show costs rise most steeply with intelligence, while output tokens barely predict the bill.

    Video from @arena's post

Heat Trend

Not enough continuous observations to show a trend yet.