Skip to content
Trending storyDeveloping

Grok Voice 3.0 tops Design Arena's AudioAgentBench at 85.1%

2 articles1 sourcesince Oct 15Last article 2h ago ·

Overview

AISummary of 2 articles

Design Arena says Grok Voice 3.0, built by xAI, ranks first on AudioAgentBench with 85.1%, 7.4 points above Grok Voice Think Fast 2.0 and ahead of OpenAI's GPT Live 1.

The ranking was announced on X on October 15, 2026.

AudioAgentBench is a Design Arena leaderboard for speech-to-speech models acting as voice agents. Design Arena says it tests six multi-turn, customer-facing scenarios such as booking appointments, placing grocery orders and planning events, and scores tool use, instruction following and state tracking. It uses recordings of real human speakers rather than synthetic voices.

The ranking comes from Design Arena's own benchmark, which it created and launched the same day.

Written by AI from the articles below · updated Oct 15, 1:26 PM ET

Check the sources:

Latest developmentOct 15, 1:14 PM ET
Design Arena launches AudioAgentBench to rank speech-to-speech voice agents

Developments

2 developments

  1. Oct 15, 1:14 PM ET · 1 article
    Design Arena launches AudioAgentBench, a speech-to-speech voice agent leaderboard
    Design Arena: Design Arena launches AudioAgentBench to rank speech-to-speech voice agents
  2. Oct 15, 1:04 PM ET · 1 article
    Grok Voice 3.0 tops Design Arena AudioAgentBench speech-to-speech leaderboard at 85.1%
    Design Arena: Grok Voice 3.0 tops Design Arena's AudioAgentBench at 85.1%

Article timeline

The articles in this story. Times are ET.

Oct 15
  1. Design ArenaOfficial
    Design Arena launches AudioAgentBench to rank speech-to-speech voice agents

    AIDesign Arena has launched AudioAgentBench, a leaderboard that tests speech-to-speech models as voice agents across six multi-turn, customer-facing scenarios such as booking appointments, placing grocery orders and planning events. The benchmark scores tool use, instruction following and state tracking, using recordings of real human speakers rather than synthetic voices.

  2. Design ArenaOfficial
    Grok Voice 3.0 tops Design Arena's AudioAgentBench at 85.1%

    AIDesign Arena says Grok Voice 3.0 ranks first on AudioAgentBench, its speech-to-speech leaderboard, with a score of 85.1%. That is 7.4 points above Grok Voice Think Fast 2.0, the previous leader, and ahead of OpenAI's GPT Live 1.

    Image from @DesignArena's post

Heat trend

Not enough continuous observations to show a trend yet.