Skip to content
Trending storyDeveloping

Arena says OpenAI models hold six of 11 Agent Arena Pareto frontier spots

1 article1 sourcesince Oct 20Last article 2h ago ·

Overview

AISummary of 1 article

Arena says OpenAI's GPT-6 Luna and GPT-6.1 Sol hold six of the 11 spots on its Agent Arena Pareto frontier across their reasoning efforts.

Arena says GPT-6 Luna takes three of the four frontier spots below $0.10 per task, using Max, Medium and Low reasoning. Arena says GPT-6.1 Sol sets the frontier between $0.22 and $0.65 per task, with the same three reasoning levels on it. Arena adds that GPT-6 Astra (Max) is also on the frontier, which would give OpenAI seven of the 11 spots.

Written by AI from the articles below · updated Oct 20, 1:57 PM ET

Check the sources:

Latest developmentOct 20, 1:42 PM ET
OpenAI models hold six of 11 Agent Arena Pareto frontier spots

Article timeline

The articles in this story. Times are ET.

Oct 20
  1. Arena.aiOfficial
    OpenAI models hold six of 11 Agent Arena Pareto frontier spots

    AIOpenAI's GPT-6 Luna and GPT-6.1 Sol hold six of the 11 Agent Arena Pareto frontier spots across their reasoning efforts. GPT-6 Luna claims three of the four optimal spots below $0.10 per task with Max, Medium and Low reasoning, while GPT-6.1 Sol sets the frontier between $0.22 and $0.65 per task. Arena says that with GPT-6 Astra (Max) also on the frontier, OpenAI models hold seven of the 11 spots.

    Image from @arena's post

Heat Trend

Not enough continuous observations to show a trend yet.