Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Zoom / Scribe

Scribe v2 Pro

Proprietary multilingual speech recognition behind Zoom’s Scribe service, with synchronous file, asynchronous batch and live WebSocket access. Backend checkpoint pinning is not documented.

Transcription
availableClosed weightsSpeech recognition
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
11 of 22 specification fields collected
Output formatsJSON transcript; display/lexical text, timestamped segments and optional speaker/word fieldsSee specification notes below
Languages11 listedFull language list in specifications
Input billingSee ratesZoom AI Services · fast
Output billingSee ratesPublished units and conditions below

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
Not collected
Maximum output
Not collected
Input modalities
AudioSource for Input modalities (opens in a new tab)

Audio tracks may be supplied in video containers; no visual understanding is implied.

Output modalities
TextSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
Not collected
Structured output
Not collected
Streaming
SupportedSource for Streaming (opens in a new tab)

Live mode returns finalized transcript segments after detected turns, not word-by-word provisional text.

Batch processing
SupportedSource for Batch processing (opens in a new tab)
Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
Not collected
Parameters
Not collected
Architecture
Not collected
Languages
English (en-US), Chinese Simplified (zh-CN), Japanese (ja-JP), Spanish (es-ES), Italian (it-IT), French (fr-FR), German (de-DE), Arabic Saudi Arabia (ar-SA), Arabic UAE (ar-AE), Portuguese Brazil (pt-BR), Portuguese Portugal (pt-PT)Source for Languages (opens in a new tab)

Eleven locale codes, not eleven distinct languages. The public playground exposes a smaller subset.

Context caching
Not collected
Benchmark model identity
zoom/scribe_v2_proSource for Benchmark model identity (opens in a new tab)

Benchmark identity, not a request selector.

Product announcement name
ASR Model Pro / Scribe ProSource for Product announcement name (opens in a new tab)
Audio formats
MP3, WAV, MP4, FLAC, OGGSource for Audio formats (opens in a new tab)

Examples of file-mode formats; Live mode requires PCM16.

diarization
SupportedSource for diarization (opens in a new tab)

Fast and Batch only; separate billable diarization rate. No hard speaker limit, fewer than 20 speakers recommended.

Word timestamps
SupportedSource for Word timestamps (opens in a new tab)

Optional word_time_offsets for file transcription; not inferred for Live.

Output formats
JSON transcript; display/lexical text, timestamped segments and optional speaker/word fieldsSource for Output formats (opens in a new tab)

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

Zoom AI Services · fast

Provider source (opens in a new tab)
API model ID
Not collected

The response example reports zoom-asr-en-v1; requests select the service route and language, not a pinned checkpoint. This is not an independently selectable model ID.

Access
POST /v2/aiservices/scribe/transcribe with a source file URL and language/config. Fast mode returns a completed JSON transcript.
Regions
Cloud service; model-specific execution regions and residency options require account confirmation
Reported response model ID
zoom-asr-en-v1Source for Reported response model ID (opens in a new tab)

Response field/example; not a request parameter or a guarantee that a particular underlying checkpoint is pinned.

Streaming
Not supportedSource for Streaming (opens in a new tab)

Completed response or asynchronous file job, not a live stream.

diarization
SupportedSource for diarization (opens in a new tab)

Fast and Batch only; separate billable diarization rate. No hard speaker limit, fewer than 20 speakers recommended.

Audio channels
Mono or stereo; optional separate channel transcriptsSource for Audio channels (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Speech recognitionfast transcription1$0.0033processed audio minute
Speech recognitionfast transcription · India price2₹0.31processed audio minute
Speech recognitionfast transcription · advertised hourly rate3$0.20processed audio hour
Speech recognitionSpeaker diarization add-on4$0.0008processed audio minute
Speech recognitionSpeaker diarization add-on · India price5₹0.07processed audio minute

Rate conditions

  1. Published Build Platform USD consumption rate. Batch is marked paid-plan only. File processing and diarization are separately priced. Do not apply SDK-session rounding to Scribe without an API-specific billing contract.
  2. Published INR storefront rate, excluding GST. Alternative regional currency price, not an additional charge or an inferred exchange-rate conversion.
  3. Advertised hourly equivalent. Fast and Live have rounding differences from the separately published per-minute credit rates. These are alternative presentations, not charges to add together.
  4. Additional rate when diarization is enabled in Fast or Batch mode. Paid-plan only. Do not interpret as the complete transcription price.
  5. Additional INR diarization charge on eligible paid plans; excluding GST. Not a currency conversion of the USD rate.
Pricing source (opens in a new tab)Pricing source (opens in a new tab)Pricing source (opens in a new tab)

Data & enterprise terms

Zoom’s terms say communications-like Customer Content is not used to train Zoom or third-party AI models. This does not establish zero retention: operational data, configured storage and service-generated data have separate handling. No API-specific numeric retention window or execution region was verified.

Read the service policy (opens in a new tab)

Zoom AI Services · batch

Provider source (opens in a new tab)
API model ID
Not collected

The response example reports zoom-asr-en-v1; requests select the service route and language, not a pinned checkpoint. This is not an independently selectable model ID.

Access
POST /v2/aiservices/scribe/jobs with S3 input/output. Supports SINGLE, PREFIX and MANIFEST; poll job/file status or receive a signed webhook.
Regions
Cloud service; model-specific execution regions and residency options require account confirmation
Reported response model ID
zoom-asr-en-v1Source for Reported response model ID (opens in a new tab)

Response field/example; not a request parameter or a guarantee that a particular underlying checkpoint is pinned.

Streaming
Not supportedSource for Streaming (opens in a new tab)

Completed response or asynchronous file job, not a live stream.

diarization
SupportedSource for diarization (opens in a new tab)

Fast and Batch only; separate billable diarization rate. No hard speaker limit, fewer than 20 speakers recommended.

Audio channels
Mono or stereo; optional separate channel transcriptsSource for Audio channels (opens in a new tab)
Batch limits
Maximum 1,000 files per manifest; single-file and S3-prefix jobs also supportedSource for Batch limits (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Speech recognitionbatch transcription1$0.0025processed audio minute
Speech recognitionbatch transcription · India price2₹0.23processed audio minute
Speech recognitionbatch transcription · advertised hourly rate3$0.15processed audio hour
Speech recognitionSpeaker diarization add-on4$0.0008processed audio minute
Speech recognitionSpeaker diarization add-on · India price5₹0.07processed audio minute

Rate conditions

  1. Published Build Platform USD consumption rate. Batch is marked paid-plan only. File processing and diarization are separately priced. Do not apply SDK-session rounding to Scribe without an API-specific billing contract.
  2. Published INR storefront rate, excluding GST. Alternative regional currency price, not an additional charge or an inferred exchange-rate conversion.
  3. Advertised hourly equivalent. Fast and Live have rounding differences from the separately published per-minute credit rates. These are alternative presentations, not charges to add together.
  4. Additional rate when diarization is enabled in Fast or Batch mode. Paid-plan only. Do not interpret as the complete transcription price.
  5. Additional INR diarization charge on eligible paid plans; excluding GST. Not a currency conversion of the USD rate.
Pricing source (opens in a new tab)Pricing source (opens in a new tab)Pricing source (opens in a new tab)

Data & enterprise terms

Zoom’s terms say communications-like Customer Content is not used to train Zoom or third-party AI models. This does not establish zero retention: operational data, configured storage and service-generated data have separate handling. No API-specific numeric retention window or execution region was verified.

Read the service policy (opens in a new tab)

Zoom AI Services · live

Provider source (opens in a new tab)
API model ID
Not collected

Benchmark identifier is not a public inference selector.

Access
Backend WebSocket at wss://api.zoom.us/v2/aiservices/scribe/live; live-asr subprotocol and JWT Authorization. Use a backend proxy for browser audio.
Regions
Cloud service; model-specific execution regions and residency options require account confirmation
Streaming
SupportedSource for Streaming (opens in a new tab)

Finalized transcript segment after each speech turn.

Audio formats
Little-endian PCM16, monoSource for Audio formats (opens in a new tab)
Audio sample rates
16 kHzSource for Audio sample rates (opens in a new tab)
Streaming audio chunks
Approximately 100 ms frames recommendedSource for Streaming audio chunks (opens in a new tab)
Audio duration limits
Maximum session 120 minutes; idle timeout after 30 seconds without audioSource for Audio duration limits (opens in a new tab)
Request quotas
Per-account concurrent sessions: default 20; enterprise 100; large contact center 1,000; custom 5,000+Source for Request quotas (opens in a new tab)

Tier-specific published quotas, not every account’s guaranteed entitlement.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
Speech recognitionlive transcription1$0.0042processed audio minute
Speech recognitionlive transcription · India price2₹0.39processed audio minute
Speech recognitionlive transcription · advertised hourly rate3$0.25processed audio hour

Rate conditions

  1. Published Build Platform USD consumption rate. Batch is marked paid-plan only. File processing and diarization are separately priced. Do not apply SDK-session rounding to Scribe without an API-specific billing contract.
  2. Published INR storefront rate, excluding GST. Alternative regional currency price, not an additional charge or an inferred exchange-rate conversion.
  3. Advertised hourly equivalent. Fast and Live have rounding differences from the separately published per-minute credit rates. These are alternative presentations, not charges to add together.
Pricing source (opens in a new tab)Pricing source (opens in a new tab)Pricing source (opens in a new tab)

Data & enterprise terms

Zoom’s terms say communications-like Customer Content is not used to train Zoom or third-party AI models. This does not establish zero retention: operational data, configured storage and service-generated data have separate handling. No API-specific numeric retention window or execution region was verified.

Read the service policy (opens in a new tab)

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

Zoom AI Services · fast

Not verified

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

How it works
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Minimum reusable input
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Lifetime & refresh
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

How to enable it
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Usage fields
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Billing rules
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Limits & data boundaries
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Zoom AI Services · batch

Not verified

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

How it works
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Minimum reusable input
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Lifetime & refresh
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

How to enable it
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Usage fields
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Billing rules
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Limits & data boundaries
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Zoom AI Services · live

Not verified

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

How it works
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Minimum reusable input
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Lifetime & refresh
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

How to enable it
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Usage fields
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Billing rules
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

Limits & data boundaries
Not collected

No model-specific discounted prompt-cache contract was verified. Audio buffering, S3 job storage and orchestration memory are not treated as prompt caching.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

What the measurements say

Open ASR · LibriSpeech test-clean

Hugging Face Open ASR Leaderboard · independent benchmark
1.11% WER

Pinned official results commit d3c0e9a07b475a64afb56254e8c35b584905af32. LS Clean WER column; lower is better. Dataset-specific result, not a global quality score or a speed measurement.

Evaluation source (opens in a new tab)

Open ASR · AMI-Cleaned

Hugging Face Open ASR Leaderboard · independent benchmark
6.08% WER

Pinned official results commit d3c0e9a07b475a64afb56254e8c35b584905af32. AMI-Cleaned column, not the older AMI dataset column. Aggregate results cover different subsets across releases and are not used as comparable averages.

Evaluation source (opens in a new tab)

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

Hugging Face’s pinned Open ASR results explicitly mark zoom/scribe_v2_pro as Proprietary. Zoom calls the current Scribe backend ASR Model Pro/Scribe Pro and links to that leaderboard. Exact backend pinning and response-ID version mapping remain unverified.

Classification evidence (opens in a new tab)

Sources reviewed

Zoom: api (opens in a new tab)Reviewed 13 Oct 2026