Tasks and inputs
Scribe v2 Medical is listed for Automatic speech recognition, Speech → Text. It accepts audio, video, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
SyncDev model directory · Profile updated Nov 17, 2026
Scribe v2 Medical is ElevenLabs' HIPAA-eligible batch transcription model for clinical audio, with 35% fewer errors than Scribe v2 on medical speech.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Pricing conditions: About $0.20 per hour of audio, the same rate as Scribe v2. Indicative; confirm on your ElevenLabs plan.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
Drug names break general transcription. So do anatomy and pathology terms, and a dictating clinician speaking fast. Scribe v2 Medical is ElevenLabs' answer: a fine-tune of Scribe v2 for clinical audio, generally available since 11 September 2026. ElevenLabs reports 35% fewer transcription errors than the base model on clinical recordings, with no loss on everyday speech.
It produces draft transcripts for a qualified professional to review. It does not interpret clinical information, and it does not diagnose or suggest treatment. Treat it as a typist, not a colleague.
For most buyers this section decides everything.
Without an Enterprise BAA, do not send identifiable patient audio. Use synthetic or de-identified recordings for evaluation.
Nothing in your integration needs to change beyond the model ID. Requests go to POST /v1/speech-to-text with model_id: "scribe_v2_medical", and the same options apply: keyterms, entity detection, diarization, timestamps, no-verbatim mode and audio tagging. Languages (90+) and price match Scribe v2 too, at about $0.20 per hour of audio.
| Field | Value |
|---|---|
| Model ID | scribe_v2_medical |
| Mode | Batch only |
| Input | Audio, video |
| Languages | 90+ |
| Indicative price | $0.20 per audio hour |
| Live alternative | Scribe v2 Realtime |
There is no streaming version. For live ambient documentation during a consultation, ElevenLabs points you to Scribe v2 Realtime, which lacks the medical fine-tune. Many clinical products work fine with batch: record the encounter, transcribe after, send the draft to the clinician. If yours needs words on screen during the visit, this model is not the fit.
The general-purpose batch model has since moved on to Scribe v3, which cuts word errors substantially across languages. The medical fine-tune is built on v2. On non-clinical audio the newer general model may well beat it; on dense clinical vocabulary the specialist should still lead. Testing both is cheap.
Collect thirty de-identified recordings that match your setting: dictations, intake calls, multi-speaker consultations. Add your formulary and specialty terms as keyterms. Have a clinician mark every error by category (medication, dosage, anatomy, everything else); a wrong dose is not the same as a dropped filler word. Compare against MAI-Transcribe-2 or other speech-to-text models on the same set.
Compare the model on the work you need to complete.
Scribe v2 Medical is listed for Automatic speech recognition, Speech → Text. It accepts audio, video, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.
Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
ElevenLabs
Scribe v3 is ElevenLabs' batch speech-to-text model, cutting word errors about 27% versus Scribe v2 across 102 languages, at $0.20 per audio hour.
inclusionAI
Ming-3.0-asr-tiny is inclusionAI's open speech-recognition model for Chinese and English with hotword hints, dialect coverage and context-aware ASR.