Groq (Whisper endpoints)
Groq's LPU-based Whisper-large-v3 endpoint — exceptionally low-latency transcription.
Groq's LPU-based Whisper-large-v3 endpoint — exceptionally low-latency transcription.
Best for latency-sensitive transcription where Groq's LPU throughput is the differentiator. Pricing: from $0.111/hr (whisper-large-v3, batch).
What it is
Groq hosts Whisper-large-v3 and Whisper-large-v3-turbo on its LPU inference hardware via an OpenAI-compatible /audio/transcriptions endpoint. Strong selling point is real-wall-clock speed — transcribing audio many times faster than its duration. No streaming, no built-in diarization. Pricing is per-hour-of-audio. Good fit for product flows where users wait for a transcript to render. Best fit: latency-sensitive transcription where groq's lpu throughput is the differentiator. Caveats: batch only at last check; no streaming endpoint; no built-in diarization. Pricing as listed: from $0.111/hr (whisper-large-v3, batch). Feature flags from vendor docs: word-level timestamps. Directory tags: commercial-api, model-host. Last vendor-page check: 2026-05-12.
Watch out for: Batch only at last check; no streaming endpoint; no built-in diarization.
Install / use
POST https://api.groq.com/openai/v1/audio/transcriptions (model=whisper-large-v3)Features
| Speaker diarization | No |
| Word-level timestamps | Yes |
| Streaming / real-time | No |
| Languages supported | 99 |
| HIPAA eligible | No |
Groq (Whisper endpoints) vs Whipscribe
| Feature | Groq (Whisper endpoints) | Whipscribe |
|---|---|---|
| Category | Transcription APIs | Transcription APIs |
| Pricing | from $0.111/hr (whisper-large-v3, batch) | free beta |
| Speaker diarization | No | Yes |
| Word timestamps | Yes | Yes |
| Streaming | No | No |
| Languages | 99 | 99 |
| Platforms | API | Web, API, MCP |
Alternatives to Groq (Whisper endpoints)
Whipscribe is a managed faster-whisper + whisperX service. If you want transcripts without running infrastructure, paste a URL or drop a file in the form below — you'll have a transcript in seconds.