ModelCards
← All benchmarks

Short phrases (19 languages)

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 Transcribe 2.0 · xAI
    6.8 % WER

“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 Transcribe 2.0 launch post

    xAI's internal short-phrase set (voice-assistant utterances, 19 languages), word error rate, lower is better. The post's other WER charts (telephony, conversational, credentials) give no numbers in text and are not reproduced.

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