Grok Voice Transcribe 2.0 logo

Grok Voice Transcribe 2.0

SpaceXAI · Released Sep 2026

Conditional

Grok Voice Transcribe 2.0 is SpaceXAI's speech-to-text model at $0.10 per hour of audio in batch, with speaker labels, word timestamps and key terms included. Conditional because every accuracy figure is SpaceXAI's own: it says the model is twice as accurate as 1.0 and first among 32 streaming models on Artificial Analysis, and we could not check that leaderboard. Right for voice-agent and call-transcription developers who will test it on their own audio; we have not tested it.

Is it right for you?

Good for

  • Streaming transcription for voice agents: transcribes an audio stream in real time at $0.20 per hour and offers smart turn detection to find the end of a speaker's turn
  • Call and meeting audio that needs structure: speaker labels at no additional cost, word-level start and end times with confidence scores, and up to 8 channels transcribed independently
  • Domain vocabulary: up to 100 key terms per request, such as product names or medical vocabulary
  • Cheap file transcription: batch is $0.10 per hour of audio, with diarization, timestamps and key terms included
  • Multilingual audio: transcribes dozens of languages, detects the language automatically and follows mid-recording switches in a single pass

Not good for

  • Decisions that rest on the accuracy claims: the twice-as-accurate-as-1.0 figure, the telephony result and the first place among 32 streaming models are all SpaceXAI's own, and we could not read the leaderboard scores
  • Pinning an API model id today: the launch post gives no id for 2.0 and says only that 2.0 will soon be the default and that `grok-voice-transcribe-1.0` can be pinned to stay on 1.0
  • Staying on 1.0: SpaceXAI says 1.0 will be deprecated in the coming weeks, so plan the move to 2.0

Pricing

Input

$0.10 / hour of audio (batch)

Output

Included in per-hour audio rate

Context

Streaming $0.20 / hour of audio

View full pricing

Benchmarks

BenchmarkScoreSource
Artificial Analysis streaming accuracy rank (vendor-reported)1st of 32 streaming models Source
Short-phrase word error rate, 1.0 to 2.0 (vendor-reported)20.6% to 6.8% Source

No verdict changes yet

The clock starts day one. Changes land here as our verdict evolves.

Verification log

  • Pricing· No changes

    Automated agent

  • Pricing· No changes

    Automated agent

How we evaluate