Grok Voice Transcribe 2.0
Grok Voice Transcribe 2.0 is SpaceXAI's current speech-to-text model. Use the explicit model ID grok-voice-transcribe-2.0 for REST file transcription or WebSocket streaming; the live docs price them at $0.10 and $0.20 per audio hour.
The release adds a low-cost transcription route with word timestamps, speaker and channel separation, domain keyterms, and live end-of-turn detection. Pinning the model matters because SpaceXAI's live reference pages name 2.0 as the default while the September 17 release note still names 1.0.
SpaceXAI's speech-to-text model page and capability guide control current pricing, formats, endpoints, and features. The release note confirms the 2.0 launch but conflicts on the default. SpaceXAI's accuracy comparison against 1.0 remains a provider claim rather than an independent benchmark.
- Transcribe interviews and meetings.
- Separate call-center channels or speakers.
- Generate live captions and dictation.
- Bias recognition toward product and domain terms.
- Detect turn completion in streaming voice workflows.
The REST /v1/stt endpoint accepts a file or server-downloadable URL and supports files up to 500 MB. The WebSocket endpoint streams raw audio and can return interim and final transcripts. REST is $0.10 per hour and streaming is $0.20 per hour.
The response can include word-level timestamps and detected language. Streaming and supported REST options add diarization, up to eight interleaved channels, keyterm biasing, filler-word control, voice-activity thresholds, endpointing, and Smart Turn.
The live model and capability pages say 2.0 is the default, but the September 17 release note says 1.0. Until SpaceXAI reconciles them, pass grok-voice-transcribe-2.0 explicitly and record the selected model in your own logs.
Grok Voice Transcribe 2.0 FAQ
Common questions about Grok Voice Transcribe 2.0.
What model ID should I use for Grok Voice Transcribe 2.0?+
Pass grok-voice-transcribe-2.0 explicitly. First-party pages disagree about the omitted-model default, so explicit selection is safer.
How much does Grok speech to text cost?+
SpaceXAI lists REST transcription at $0.10 per audio hour and streaming transcription at $0.20 per audio hour.
Does it support speaker diarization and multichannel audio?+
Yes. The current guide documents speaker IDs plus separate transcripts for up to eight interleaved channels, with an Opus limitation for multichannel streaming.