Grok Voice Transcribe 2.0 tops streaming ASR accuracy with 2.7% WER at same price
ArtificialAnlys · x · 2026-09-19
Artificial Analysis evaluated SpaceXAI's new speech-to-text model Grok Voice Transcribe 2.0, which takes #1 on the AA-WER Streaming leaderboard.
- Streaming final transcript: WER improved from 3.9% to 2.7% at 0.49s after end of speech — more accurate than Muse Voice Transcribe (3.1%/0.16s), ElevenLabs Scribe v2 Realtime (3.6%/0.14s), and Cartesia Ink-2 (3.4%/0.43s), though slightly slower
- First partial transcript: 3.4% WER at 0.49s, ahead on accuracy but trailing rivals' 0.13-0.17s latency
- Non-streaming: 2.3% AA-WER, ranking #5 of 59 models, behind Alibaba's Fun-Realtime-ASR-preview and StepFun's StepAudio 3 ASR (both 1.7%), Microsoft MAI-Transcribe-2 (2.0%), and ElevenLabs Scribe v2 (2.2%)
- Pricing unchanged: $0.20/hour streaming ($3.33 per 1,000 minutes), roughly half of ElevenLabs Scribe v2 Realtime and Deepgram Flux ($6.50); non-streaming $1.67 per 1,000 minutes
Available now via API in both streaming and non-streaming modes.
Related event: xAI Launches Grok Voice Transcribe 2.0, Tops Streaming ASR Leaderboard(15 posts)→
More from Models
- Fable 5.5 delayed as Anthropic reportedly bets OpenAI's Astra 6.2 won't beat it — haider1 · 2026-11-04
- Community asks Moonshot when Kimi K3.1 model weights will drop — bgmshana · 2026-11-04
- Meta's Muse Spark Open license counts distilled models as derivatives — aidenclarke_12 · 2026-11-04
- Tencent Hy4 trails GPT-6.1 on hard tests as Kimi K3.1 Pro beats it on two benchmarks — heyshrutimishra · 2026-11-04
- Tencent's Hy4 agent self-trained a smaller model that beat Tencent's own baseline — heyshrutimishra · 2026-11-04
- Hy4 tops open models in finance eval at 3.04, hallucination rate cut to 5.45% — heyshrutimishra · 2026-11-04