ModelCards
← All benchmarks

Speech Agent Arena

We haven't written a description for this benchmark yet; see the card notes below for how each maker ran it.

Your workloadRAG over docs · 12K in · 600 out · 50K requests/mo · 30% cached
  1. 1Grok Voice 3.0 · xAI
    1070 Elo

“Best value”: no other model here reports a higher score at a lower cost for your workload. Scores are self-reported by each maker with different settings, so treat close calls as ties.

How each maker reported it

  • Grok Voice 3.0 launch post (Artificial Analysis, 13 Oct 2026)

    Third-party results as published by xAI in its launch post, not verified here. Competitor configurations as labelled by xAI: GPT-Live-1 (Astra, Medium), Gemini 3.8 Live (Ext. Thinking, High). xAI also reports 71.6% on τ-Voice and 74% accuracy on ServiceNow's EVA-Bench.

Only each model's own maker's number is shown; scores a maker reports for rival models are left out.