AI Services API

Content-Type: application/json

Example:

{
  "vocabularies": [
    {
      "vocabulary_id": "acct_vocab_abc123",
      "language": "en-US",
      "name": "vocabulary name"
    }
  ],
  "next_page_token": "eyJqb2JfaWQiOiJqb2JfZGVmNDU2In0"
}

• language

  string — The language code. For details, refer to [Abbreviations](https://developers.zoom.us/docs/api/references/abbreviations/).

• name

  string | null — The name of the vocabulary.

• phrases

  array — A list of phrases used to improve recognition of domain-specific terms, names, and other uncommon vocabulary. This array supports a maximum of 100 items.

  Items:

  string

Content-Type: application/json

Example:

{
  "vocabulary_id": "acct_vocab_abc123",
  "language": "en-US",
  "phrases": [
    "Acme"
  ],
  "name": "vocabulary name"
}

Content-Type: application/json

Example:

{
  "vocabulary_id": "acct_vocab_abc123",
  "language": "en-US",
  "phrases": [
    "Acme"
  ],
  "name": "vocabulary name"
}

• input (required)

  object — Shared input evaluated against every question: text or one or more user messages.

• questions (required)

  array — One to ten independent typed questions. Names must be unique within this request.

  Items:

  One of:

  • instructions (required)

    string — Instructions describing the decision to make. Input must not override service-level instructions or guardrails.

  • name (required)

    string — Caller-defined question name. Names must be unique within this request; the corresponding answer echoes the name.

  • type (required)

    string, possible values: "predicate" — The decision question type.

  • choices (required)

    array — Between 2 and 255 allowed choices. Choice values, including booleans, must be unique within this question.

    Items:

    • value (required)

      object — A string or boolean choice value. Values must be unique within a question.

    • description

      string — Optional guidance describing when this choice applies.

  • instructions (required)

    string — Instructions describing the decision to make. Input must not override service-level instructions or guardrails.

  • name (required)

    string — Caller-defined question name. Names must be unique within this request; the corresponding answer echoes the name.

  • type (required)

    string, possible values: "choice" — The decision question type.

  • instructions (required)

    string — Instructions describing the decision to make. Input must not override service-level instructions or guardrails.

  • levels (required)

    array — Between 2 and 10 ordered scoring levels, indexed from zero.

    Items:

    • label (required)

      string — Human-readable level name.

    • description

      string — Optional guidance describing the level.

  • name (required)

    string — Caller-defined question name. Names must be unique within this request; the corresponding answer echoes the name.

  • type (required)

    string, possible values: "score" — The decision question type.

• enable_image_input

  boolean, default: false — Optional. Defaults to `false`. Set to `true` to opt in to image input when the input contains an `input_image`.

• model

  string, possible values: "zoom-decisions", "zoom-decisions-pro", default: "zoom-decisions" — Optional decision model ID. Currently supported IDs are `zoom-decisions` and `zoom-decisions-pro`. When omitted, the service uses `zoom-decisions`. Image input is opt-in: set `enable_image_input` to `true`. Selecting `zoom-decisions-pro` does not require additional entitlement.

• reference_id

  string — Optional caller-provided correlation and tracing identifier.

Content-Type: application/json

Example:

{
  "answers": [
    {
      "type": "predicate",
      "name": "damaged",
      "probability": 0.9964802288772219
    }
  ],
  "model": "zoom-decisions",
  "request_id": "req_test_case_predicate",
  "usage": {
    "input_units": 96,
    "output_units": 1,
    "unit_type": "tokens",
    "details": {}
  }
}

Content-Type: application/json

Example:

{
  "jobs": [
    {
      "job_id": "job_abc123",
      "state": "PROCESSING",
      "submitted_at": "2026-01-05T18:34:12Z",
      "input": {
        "mode": "INPUT_PREFIX",
        "source": "S3",
        "uri": "s3://acme-bucket/audio/"
      },
      "output": {
        "destination": "S3",
        "uri": "s3://acme-bucket/transcripts/",
        "layout": "PREFIX"
      },
      "config": {
        "language": "en-US",
        "timestamps": true,
        "word_time_offsets": false,
        "channel_separation": false,
        "diarization": false,
        "profanity_filter": false,
        "output_format": "json"
      },
      "summary": {
        "total_files": 2487,
        "queued": 0,
        "processing": 23,
        "succeeded": 2450,
        "failed": 14,
        "skipped": 0,
        "canceled": 0
      },
      "error": {
        "code": "INVALID_CREDENTIALS",
        "message": "AWS credentials expired",
        "details": {
          "additionalProperty": ""
        }
      }
    }
  ],
  "next_page_token": "eyJqb2JfaWQiOiJqb2JfZGVmNDU2In0"
}

• config (required)

  object — The ASR transcription configuration.

  • channel_separation

    boolean, default: false — Whether to transcribe stereo channels separately.

  • diarization

    boolean — The stretch or premium feature for speaker diarization.

  • language

    string — The language code. See following page for details: https://developers.zoom.us/docs/api/references/abbreviations/

  • language_detection

    object — Zoom Scribe supports optional automatic language detection through `config.language_detection`.

    • candidate_languages

      array | null, default: ["en-US"] — An optional set of languages to consider during language detection.

    • mode

      string, possible values: "once", "continuous", default: "continuous" — The language detection mode. Supported values are `once` and `continuous`.

  • model

    string | null — The optional model selection.

  • output_format

    string — The output format. The default is JSON. Reserved for future (srt/vtt).

  • profanity_filter

    boolean — Whether to enable profanity filtering.

  • segmentation_mode

    string, possible values: "auto", "none" — The VAD option. The default value is 'auto'. auto: use default VAD behavior, none: disable VAD segmentation.

  • vocabulary

    object — The vocabulary used to customize speech recognition for the current request. This field takes higher precedence over `vocabulary_id` if both are present.

    • phrases

      array — A list of phrases used for ephemeral vocabulary.

      Items:

      string — A phrase used for ephemeral vocabulary.

  • vocabulary_id

    string — A reference to an immutable vocabulary asset used to customize speech recognition.

  • word_time_offsets

    boolean, default: false — The include word-level timestamps.

• input (required)

  object — The input source configuration. Either 'uri' (for single/prefix modes) or 'manifest' (for manifest mode) must be provided.

  • source (required)

    string, possible values: "S3" — The storage provider.

  • auth

    object — The authentication credentials for the input source.

    • aws

      object — The AWS credentials for S3 access. Required when source is 's3' and URIs are not pre-signed. Use session tokens for enhanced security.

      • access_key_id (required)

        string — The AWS access key ID.

      • secret_access_key (required)

        string — The AWS secret access key.

      • session_token (required)

        string — The AWS session token for temporary credentials (recommended for security).

  • filters

    object — The file filtering options for prefix mode. Use glob patterns to include/exclude files from processing.

    • exclude_globs

      array — The exclude files that match the glob patterns. Applied after include_globs.

      Items:

      string

    • include_globs

      array — The include only files that match the glob patterns. Supports wildcards: * (any chars), ** (recursive), ? (single char). Applied first.

      Items:

      string

  • manifest

    array — The explicit list of file URIs for manifest mode. Each URI can include SAS tokens or signed URLs. Required when mode is 'manifest'. Maximum 1000 files per batch job.

    Items:

    string, format: uri

  • mode

    string, possible values: "SINGLE", "PREFIX", "MANIFEST" — The input processing mode. Optional - can be inferred: 'SINGLE' if uri points to single file, 'PREFIX' if uri is directory/prefix, 'MANIFEST' if manifest array provided.

  • uri

    string, format: uri — The S3 URI Blob URL for single file or prefix/container. Required for single and prefix modes.

• output (required)

  object — The output destination configuration.

  • destination (required)

    string, possible values: "S3" — The output storage provider.

  • uri (required)

    string, format: uri — The output base URI for a single transcript file or a path for multiple transcript files.

  • auth

    object — The authentication credentials for the output destination.

    • aws

      object — The AWS credentials for S3 access. Required when source is 's3' and URIs are not pre-signed. Use session tokens for enhanced security.

      • access_key_id (required)

        string — The AWS access key ID.

      • secret_access_key (required)

        string — The AWS secret access key.

      • session_token (required)

        string — The AWS session token for temporary credentials (recommended for security).

  • layout

    string, possible values: "SINGLE", "PREFIX", "ADJACENT" — The output file layout strategy. 'SINGLE': use 'uri' for output file (single file mode). 'PREFIX': use output uri as base directory. 'ADJACENT': place output next to input file (same directory).

• notifications

  object — The webhook notification settings for job status updates.

  • webhook_url (required)

    string, format: uri — The HTTPS webhook URL for job status notifications. It will receive POST requests on job state changes.

  • secret

    string — The HMAC secret for webhook signature verification. Used to sign notification payloads.

• outputs

  object — The optional output configuration for the transcription. This plural `outputs` object is separate from the singular `output` storage destination.

  • text_polishing

    object — The text polishing configuration.

    • enabled

      boolean, default: false — Whether to return polished text under `outputs.text_polishing.text`. Text polishing must support the requested locale.

• reference_id

  string — The customer-provided reference ID for tracking and correlation.

Content-Type: application/json

Example:

{
  "job_id": "job_abc123",
  "state": "QUEUED",
  "submitted_at": "2026-01-05T18:34:12Z"
}

Content-Type: application/json

Example:

{
  "job_id": "job_abc123",
  "state": "PARTIAL",
  "submitted_at": "2026-01-05T18:34:12Z",
  "summary": {
    "total_files": 2487,
    "queued": 0,
    "processing": 23,
    "succeeded": 2450,
    "failed": 14,
    "skipped": 0
  }
}

Content-Type: application/json

Example:

{
  "files": [
    {
      "file_id": "f_001",
      "input_uri": "s3://acme-bucket/audio/2025/11/call_0001.wav",
      "state": "FILE_SUCCEEDED",
      "output_uri": "s3://acme-bucket/transcripts/2025/11/call_0001.json",
      "duration_sec": 3520.2,
      "error": {
        "code": "UNSUPPORTED_MEDIA",
        "message": "OGG not enabled for MVP"
      }
    }
  ],
  "next_page_token": "eyJ0YXNrX2lkIjoidGFza18wMDIifQ"
}

Content-Type: application/json

Example:

{
  "file_id": "f_001",
  "input_uri": "s3://acme-bucket/audio/2025/11/meeting_001.mp3",
  "state": "FILE_SUCCEEDED",
  "output_uri": "s3://acme-bucket/transcripts/2025/11/meeting_001.json",
  "duration_sec": 183.5,
  "error": {
    "code": "UNSUPPORTED_CONTENT_TYPE",
    "message": "Content type application/pdf is not supported for transcription",
    "details": {
      "additionalProperty": ""
    }
  }
}

• config (required)

  object — The ASR configuration options.

  • channel_separation

    boolean, default: false — Whether to transcribe stereo channels separately.

  • diarization

    boolean — A stretch/premium feature for speaker diarization.

  • language

    string — The language code. For details, see the [Abbreviations page](https://developers.zoom.us/docs/api/references/abbreviations/).

  • language_detection

    object — The optional automatic language detection configuration.

    • candidate_languages

      array | null, default: ["en-US"] — An optional set of languages to consider during language detection.

    • mode

      string, possible values: "once", "continuous", default: "continuous" — The language detection mode. Supported values are `once` and `continuous`.

  • model

    string | null — The optional model selection.

  • output_format

    string — The output format. The default value is `json`. Reserved for future use (`srt`/`vtt`).

  • profanity_filter

    boolean — Whether to enable profanity filtering.

  • timestamps

    boolean, default: true — Whether to include segment-level timestamps.

  • vocabulary

    object — The vocabulary used to customize speech recognition for the current request. This field takes higher precedence over the vocabulary ID if both are present.

    • phrases

      array — A list of phrases used for ephemeral vocabulary.

      Items:

      string — A phrase used for ephemeral vocabulary.

  • vocabulary_id

    string — A reference to an immutable vocabulary asset used to customize speech recognition.

  • word_time_offsets

    boolean, default: false — Whether to include word-level timestamps.

• file (required)

  string — The audio file to transcribe. Supported formats: wav, mp3, m4a, flac, ogg, webm, mp4, mov. Max size: 100MB. Max duration: 2 hours.

• outputs

  object — The optional output configuration for the transcription.

  • text_polishing

    object — The text polishing configuration.

    • enabled

      boolean, default: false — Whether to return polished text under `outputs.text_polishing.text`. Text polishing must support the requested locale.

Content-Type: application/json

Example:

{
  "request_id": "req_123",
  "duration_sec": 27.4,
  "result": {
    "text_display": "Hello and welcome to the meeting...",
    "segments": [
      {
        "start": 0,
        "end": 3.24,
        "text": "Hello and welcome...",
        "channel": 0,
        "speaker": "speaker_1",
        "words": [
          {
            "word": "Hello",
            "start": 0,
            "end": 0.2
          }
        ]
      }
    ]
  },
  "model": "zoom-asr-en-v1",
  "outputs": {
    "text_polishing": {
      "text": "Hello, and welcome to the meeting. "
    }
  }
}

• file (required)

  string — The audio or video file to process. **Supported formats:** - Audio: WAV, MP3, M4A, FLAC, OGG, WebM - Video: MP4, MOV **Requirements:** - Maximum file size: 50 MB - Maximum duration: 2 hours - Sample rate: ≥8 kHz (16 kHz or higher recommended) - Channels: Mono or stereo

• config

  object — Configuration for the transcription and summarization parameters.

  • summary

    object — Summary generation configuration options.

    • citations

      boolean, default: true — Whether to include citations linking summary content back to the transcript.

    • custom_template

      string, default: "" — Custom template for the summary in Markdown format.

    • enable_dynamic_template

      boolean, default: true — Whether to allow dynamic template selection instead of a fixed `template_id`.

    • language

      string, default: "en-US" — The language code for the generated summary.

    • pii_redaction

      boolean, default: false — Whether to redact personally identifiable information (PII) from the summary.

    • should_remove_sentiment

      boolean, default: false — Whether to remove sentiment-related language from the summary.

    • summary_length

      string, possible values: "short", "long", default: "short" — The desired length of the generated summary.

    • template_id

      string, possible values: "one_on_one", "stand_up", "brainstorm", "project_update", "discussion", "q_a", "customer_success", "user_feedback", "lecture_summary", "next_steps" — The summary template to use.

  • transcribe

    object — Transcription (ASR) configuration options.

    • channel_separation

      boolean, default: false — Whether to transcribe stereo channels separately.

    • diarization

      boolean, default: false — Whether to enable speaker diarization.

    • language

      string, default: "en-US" — The language code. See the [Zoom abbreviations](https://developers.zoom.us/docs/api/references/abbreviations/) page for details.

    • model

      string, possible values: "zoom-scribe-en", default: "zoom-scribe-en" — The ASR model identifier to use for transcription.

    • output_format

      string, possible values: "json", default: "json" — The transcript output format.

    • profanity_filter

      boolean, default: false — Whether to enable profanity filtering.

    • word_time_offsets

      boolean, default: false — Whether to include word-level timestamps in the transcript.

Content-Type: application/json

Example:

{}

Content-Type: application/json

Example:

{
  "jobs": [
    {
      "job_id": "job_abc123",
      "state": "PROCESSING",
      "submitted_at": "2026-01-05T18:34:12Z",
      "input": {
        "mode": "PREFIX",
        "source": "S3",
        "uri": "s3://acme-bucket/transcripts/"
      },
      "output": {
        "destination": "S3",
        "uri": "s3://acme-bucket/summaries/",
        "layout": "PREFIX"
      },
      "config": {
        "summary_type": "CONVERSATION",
        "task": "full_summary",
        "language": "en-US",
        "output_format": "text"
      },
      "summary": {
        "total_files": 2487,
        "queued": 0,
        "processing": 23,
        "succeeded": 2450,
        "failed": 14,
        "skipped": 0,
        "canceled": 0
      },
      "error": {
        "code": "INVALID_CREDENTIALS",
        "message": "AWS credentials expired"
      },
      "usage": {
        "input_units": 985430,
        "output_units": 62140,
        "unit_type": "characters"
      }
    }
  ],
  "next_page_token": "eyJqb2JfaWQiOiJqb2JfZGVmNDU2In0"
}

• input (required)

  object — The input source configuration. Provide `uri` (for SINGLE/PREFIX modes) or `manifest` (for MANIFEST mode).

  • source (required)

    string, possible values: "S3" — The storage provider.

  • auth

    object — The storage authentication credentials.

    • aws

      object — The AWS credentials for S3 access. Required when the source is S3 and URIs are not pre-signed.

      • access_key_id (required)

        string — The AWS access key ID.

      • secret_access_key (required)

        string — The AWS secret access key.

      • session_token (required)

        string — The AWS session token for temporary credentials (recommended).

  • filters

    object — The file filtering options for PREFIX mode. Use glob patterns to include or exclude files.

    • exclude_globs

      array — The exclude files matching these glob patterns. Applied after `include_globs`.

      Items:

      string

    • include_globs

      array — The include only files matching these glob patterns. Supports wildcards: `*` (any characters), `**` (recursive), `?` (single character). Applied before `exclude_globs`.

      Items:

      string

  • manifest

    array — The explicit list of file URIs for MANIFEST mode. Required when mode is `MANIFEST`. Maximum of 1,000 files per batch job.

    Items:

    string, format: uri

  • mode

    string, possible values: "SINGLE", "PREFIX", "MANIFEST" — The input processing mode. Optional — can be inferred: `MANIFEST` if a manifest array is provided, `SINGLE` if the URI points to a single file, or `PREFIX` if the URI is a directory/prefix.

  • uri

    string, format: uri — The S3 URI for a single file or prefix/container. Required for SINGLE and PREFIX modes.

• output (required)

  object — The output destination configuration.

  • destination (required)

    string, possible values: "S3" — The output storage provider.

  • uri (required)

    string, format: uri — The base output URI where summarization results will be written.

  • auth

    object — The storage authentication credentials.

    • aws

      object — The AWS credentials for S3 write access.

      • access_key_id (required)

        string — The AWS access key ID.

      • secret_access_key (required)

        string — The AWS secret access key.

      • session_token (required)

        string — The AWS session token for temporary credentials (recommended).

  • layout

    string, possible values: "ADJACENT", "PREFIX" — The output file layout strategy. `ADJACENT`: places output next to the input file. `PREFIX`: uses the output URI as the base directory.

  • overwrite

    boolean, default: false — Whether to overwrite existing output files.

• config

  object — The summarization configuration.

  • language

    string — The output language as an ISO locale code. Supported: `en-US`, `zh-CN`, `zh-Hans`, `ja-JP`, `es-ES`, `fr-FR`, `de-DE`, `pt-PT`, `pt-BR`, `it-IT`, `ar-SA`, `ar-AE`. Defaults to `en-US`.

  • output_format

    string, possible values: "text", default: "text" — The output format. Defaults to `text`. Currently only `text` is supported.

  • summary_type

    string, possible values: "CONVERSATION", default: "CONVERSATION" — The type of content being summarized. Defaults to `CONVERSATION`.

  • task

    string, possible values: "full_summary", "recap", "action_items", "summary", default: "full_summary" — The summarization task type. Defaults to `full_summary` if not specified. Accepted values are `full_summary`, `recap`, `action_items`, and `summary`.

• limits

  object — The job processing limits.

  • max_files

    integer — The maximum number of files to process. Excess files are skipped.

• notifications

  object — The webhook notification settings for job status updates.

  • audience

    string — The JWT audience (`aud` claim) for the signed webhook token. Required when `security_type` is `WEBHOOK_SECURITY_JWT`.

  • secret

    string — The shared secret used to sign webhook payloads. Used as the HMAC key when `security_type` is `WEBHOOK_SECURITY_HMAC`, or as the JWT signing key when `security_type` is `WEBHOOK_SECURITY_JWT`.

  • security_type

    string, possible values: "WEBHOOK_SECURITY_HMAC", "WEBHOOK_SECURITY_JWT", default: "WEBHOOK_SECURITY_HMAC" — The webhook payload signature scheme. Defaults to `WEBHOOK_SECURITY_HMAC`.

  • webhook_url

    string, format: uri — The HTTPS webhook URL to receive POST notifications on job state changes.

• reference_id

  string — The customer-provided reference ID for deduplication and correlation. Submitting the same `reference_id` twice returns the existing job instead of creating a duplicate.

Content-Type: application/json

Example:

{
  "job_id": "job_abc123",
  "state": "QUEUED",
  "submitted_at": "2026-01-05T18:34:12Z"
}

Content-Type: application/json

Example:

{
  "job_id": "job_abc123",
  "state": "PROCESSING",
  "submitted_at": "2026-01-05T18:34:12Z",
  "input": {
    "mode": "PREFIX",
    "source": "S3",
    "uri": "s3://acme-bucket/transcripts/2025/11/"
  },
  "output": {
    "destination": "S3",
    "uri": "s3://acme-bucket/summaries/2025/11/",
    "layout": "PREFIX"
  },
  "config": {
    "summary_type": "CONVERSATION",
    "task": "full_summary",
    "language": "en-US",
    "output_format": "text"
  },
  "summary": {
    "total_files": 2487,
    "queued": 0,
    "processing": 23,
    "succeeded": 2450,
    "failed": 14,
    "skipped": 0,
    "canceled": 0
  },
  "error": {
    "code": "INVALID_CREDENTIALS",
    "message": "AWS credentials expired",
    "details": {
      "additionalProperty": ""
    }
  },
  "usage": {
    "input_units": 985430,
    "output_units": 62140,
    "unit_type": "characters"
  }
}

Content-Type: application/json

Example:

{
  "files": [
    {
      "file_id": "f_001",
      "input_uri": "s3://acme-bucket/transcripts/2025/11/meeting_001.vtt",
      "state": "FILE_SUCCEEDED",
      "output_uri": "s3://acme-bucket/summaries/2025/11/meeting_001.json",
      "usage": {
        "input_units": 4820,
        "output_units": 312,
        "unit_type": "characters",
        "details": {
          "summarizer": {
            "task": "full_summary",
            "summary_type": "CONVERSATION"
          }
        }
      },
      "error": {
        "code": "UNSUPPORTED_CONTENT_TYPE",
        "message": "Content type audio/wav is not supported for summarization"
      }
    }
  ],
  "next_page_token": "eyJmaWxlX2lkIjoiZl8xMDEifQ"
}

Content-Type: application/json

Example:

{
  "file_id": "f_001",
  "input_uri": "s3://acme-bucket/transcripts/2025/11/meeting_001.vtt",
  "state": "FILE_SUCCEEDED",
  "output_uri": "s3://acme-bucket/summaries/2025/11/meeting_001.json",
  "usage": {
    "input_units": 4820,
    "output_units": 312,
    "unit_type": "characters",
    "details": {
      "summarizer": {
        "task": "full_summary",
        "summary_type": "CONVERSATION"
      }
    }
  },
  "error": {
    "code": "UNSUPPORTED_CONTENT_TYPE",
    "message": "Content type audio/wav is not supported for summarization",
    "details": {
      "additionalProperty": ""
    }
  }
}

• input (required)

  object — The input content to summarize.

  • text (required)

    string — The transcript text to summarize.

• config

  object — Summarization configuration options.

  • language

    string — The output language as an ISO locale code. Supported values: `en-US`, `zh-CN`, `zh-Hans`, `ja-JP`, `es-ES`, `fr-FR`, `de-DE`, `pt-PT`, `pt-BR`, `it-IT`, `ar-SA`, `ar-AE`. Defaults to `en-US`.

  • output_format

    string, possible values: "text", default: "text" — The output format. Defaults to `text`. Currently, only `text` is supported.

  • summary_type

    string, possible values: "CONVERSATION", default: "CONVERSATION" — The type of content being summarized. Defaults to `CONVERSATION`.

  • task

    string, possible values: "full_summary", "recap", "action_items", "summary", default: "full_summary" — The summarization task type. Defaults to `full_summary` if not specified. Accepted values: `full_summary`, `recap`, `action_items`, and `summary`.

Content-Type: application/json

Example:

{
  "request_id": "req_abc123",
  "task": "full_summary",
  "result": {
    "recap": "The team aligned on deliverables: Bob will document the auth API by Friday.",
    "summary_text": "**Authentication Module**\nBob completed the auth module. Documentation is due Friday.",
    "action_items": "- **Bob**: Document the auth API by Friday",
    "full_summary": "## Recap\nThe team aligned on deliverables: Bob will document the auth API by Friday.\n\n## Summary\n**Authentication Module**\nBob completed the auth module. Documentation is due Friday.\n\n## Action Items\n- **Bob**: Document the auth API by Friday"
  },
  "model": "zoom-summarizer-v1",
  "usage": {
    "input_units": 4820,
    "output_units": 312,
    "unit_type": "characters",
    "details": {
      "summarizer": {
        "task": "full_summary",
        "summary_type": "CONVERSATION"
      }
    }
  }
}

Content-Type: application/json

Example:

{
  "jobs": [
    {
      "job_id": "job_abc123",
      "state": "PROCESSING",
      "submitted_at": "2026-01-05T18:34:12Z",
      "input": {
        "mode": "PREFIX",
        "source": "S3",
        "uri": "s3://acme-bucket/text/"
      },
      "output": {
        "destination": "S3",
        "uri": "s3://acme-bucket/translations/",
        "layout": "PREFIX"
      },
      "config": {
        "source_language": "en-US",
        "target_languages": [
          "es-ES"
        ]
      },
      "summary": {
        "total_files": 2487,
        "queued": 0,
        "processing": 23,
        "succeeded": 2450,
        "failed": 14,
        "skipped": 0,
        "canceled": 0
      },
      "error": {
        "code": "INVALID_CREDENTIALS",
        "message": "AWS credentials expired"
      }
    }
  ],
  "next_page_token": "eyJqb2JfaWQiOiJqb2JfZGVmNDU2In0"
}

• input (required)

  object — The input source configuration. Provide `uri` (for SINGLE/PREFIX modes) or `manifest` (for MANIFEST mode).

  • source (required)

    string, possible values: "S3" — The storage provider.

  • auth

    object — The storage authentication credentials.

    • aws

      object — The AWS credentials for S3 access. Required when the source is S3 and URIs are not pre-signed.

      • access_key_id (required)

        string — The AWS access key ID.

      • secret_access_key (required)

        string — The AWS secret access key.

      • session_token (required)

        string — The AWS session token for temporary credentials (recommended).

  • filters

    object — The file filtering options for PREFIX mode. Use glob patterns to include or exclude files.

    • exclude_globs

      array — The exclude files matching these glob patterns. Applied after `include_globs`.

      Items:

      string

    • include_globs

      array — This field includes only files matching these glob patterns. Supports wildcards: `*` (any characters), `**` (recursive), `?` (single character). Applied before `exclude_globs`.

      Items:

      string

  • manifest

    array — The explicit list of file URIs for MANIFEST mode. Required when mode is `MANIFEST`. Maximum of 1,000 files per batch job.

    Items:

    string, format: uri

  • mode

    string, possible values: "SINGLE", "PREFIX", "MANIFEST" — The input processing mode. Optional — can be inferred: `MANIFEST` if a manifest array is provided, `SINGLE` if the URI points to a single file, `PREFIX` if the URI is a directory/prefix.

  • uri

    string, format: uri — The S3 URI for a single file or prefix/container. Required for SINGLE and PREFIX modes.

• output (required)

  object — The output destination configuration.

  • destination (required)

    string, possible values: "S3" — The output storage provider.

  • uri (required)

    string, format: uri — The base output URI where translation results will be written.

  • auth

    object — The storage authentication credentials.

    • aws

      object — The AWS credentials for S3 write access.

      • access_key_id (required)

        string — The AWS access key ID.

      • secret_access_key (required)

        string — The AWS secret access key.

      • session_token (required)

        string — The AWS session token for temporary credentials (recommended).

  • layout

    string, possible values: "ADJACENT", "PREFIX" — The output file layout strategy. `ADJACENT`: places output next to the input file. `PREFIX`: uses the output URI as the base directory.

  • overwrite

    boolean, default: false — Whether to overwrite existing output files.

• config

  object — The translation configuration.

  • source_language

    string — The source language as an ISO locale code. Leave empty for automatic language detection. Supported: `en-US`, `zh-CN`, `zh-Hans`, `zh-TW`, `zh-Hant`, `ja-JP`, `ko-KR`, `es-ES`, `fr-FR`, `de-DE`, `pt-PT`, `pt-BR`, `it-IT`.

  • target_languages

    array — The list of target language ISO locale codes to translate. Maximum of 1 target per request. One side of each pair must be English (`en-US`). Supported: `en-US`, `zh-CN`, `zh-Hans`, `zh-TW`, `zh-Hant`, `ja-JP`, `ko-KR`, `es-ES`, `fr-FR`, `de-DE`, `pt-PT`, `pt-BR`, `it-IT`.

    Items:

    string

• limits

  object — The job processing limits.

  • max_files

    integer — The maximum number of files to process. Excess files are skipped.

• notifications

  object — The webhook notification settings for job status updates.

  • audience

    string — The JWT audience (`aud` claim) for the signed webhook token. Required when `security_type` is `WEBHOOK_SECURITY_JWT`.

  • secret

    string — The shared secret used to sign webhook payloads. Used as the HMAC key when `security_type` is `WEBHOOK_SECURITY_HMAC`, or as the JWT signing key when `security_type` is `WEBHOOK_SECURITY_JWT`.

  • security_type

    string, possible values: "WEBHOOK_SECURITY_HMAC", "WEBHOOK_SECURITY_JWT", default: "WEBHOOK_SECURITY_HMAC" — The webhook payload signature scheme. Defaults to `WEBHOOK_SECURITY_HMAC`.

  • webhook_url

    string, format: uri — The HTTPS webhook URL to receive POST notifications on job state changes.

• reference_id

  string — The customer-provided reference ID for deduplication and correlation. Submitting the same `reference_id` twice returns the existing job instead of creating a duplicate.

Content-Type: application/json

Example:

{
  "job_id": "job_abc123",
  "state": "QUEUED",
  "submitted_at": "2026-01-05T18:34:12Z"
}

Content-Type: application/json

Example:

{
  "job_id": "job_abc123",
  "state": "PROCESSING",
  "submitted_at": "2026-01-05T18:34:12Z",
  "input": {
    "mode": "PREFIX",
    "source": "S3",
    "uri": "s3://acme-bucket/text/2025/11/"
  },
  "output": {
    "destination": "S3",
    "uri": "s3://acme-bucket/translations/2025/11/",
    "layout": "PREFIX"
  },
  "config": {
    "source_language": "en-US",
    "target_languages": [
      "es-ES"
    ]
  },
  "summary": {
    "total_files": 2487,
    "queued": 0,
    "processing": 23,
    "succeeded": 2450,
    "failed": 14,
    "skipped": 0,
    "canceled": 0
  },
  "error": {
    "code": "INVALID_CREDENTIALS",
    "message": "AWS credentials expired",
    "details": {
      "additionalProperty": ""
    }
  }
}

Content-Type: application/json

Example:

{
  "files": [
    {
      "file_id": "f_001",
      "input_uri": "s3://acme-bucket/text/2025/11/document_001.txt",
      "state": "FILE_SUCCEEDED",
      "output_uri": "s3://acme-bucket/translations/2025/11/document_001.json",
      "error": {
        "code": "UNSUPPORTED_LANGUAGE_PAIR",
        "message": "Translation between zh-CN and ja-JP is not supported"
      }
    }
  ],
  "next_page_token": "eyJmaWxlX2lkIjoiZl8xMDEifQ"
}

Content-Type: application/json

Example:

{
  "file_id": "f_001",
  "input_uri": "s3://acme-bucket/text/2025/11/document_001.txt",
  "state": "FILE_SUCCEEDED",
  "output_uri": "s3://acme-bucket/translations/2025/11/document_001.json",
  "error": {
    "code": "UNSUPPORTED_LANGUAGE_PAIR",
    "message": "Translation between zh-CN and ja-JP is not supported",
    "details": {
      "additionalProperty": ""
    }
  }
}

• text (required)

  string — The text to translate.

• config

  object — Translation configuration options.

  • source_language

    string — Source language as an ISO locale code. Leave empty for automatic language detection. Supported: `en-US`, `zh-CN`, `zh-Hans`, `zh-TW`, `zh-Hant`, `ja-JP`, `ko-KR`, `es-ES`, `fr-FR`, `de-DE`, `pt-PT`, `pt-BR`, `it-IT`.

  • target_languages

    array — List of target language ISO locale codes to translate into. Maximum of 1 target per request. One side of each pair must be English (`en-US`). Supported: `en-US`, `zh-CN`, `zh-Hans`, `zh-TW`, `zh-Hant`, `ja-JP`, `ko-KR`, `es-ES`, `fr-FR`, `de-DE`, `pt-PT`, `pt-BR`, `it-IT`.

    Items:

    string

Content-Type: application/json

Example:

{
  "request_id": "req_abc123",
  "result": {
    "translations": {
      "es-ES": "Buenos días a todos, comencemos la reunión de hoy."
    }
  },
  "model": "zoom-translator-v1",
  "usage": {
    "input_units": 62,
    "output_units": 56,
    "unit_type": "characters",
    "details": {
      "translator": {
        "source_language": "en-US",
        "target_language": "es-ES"
      }
    }
  }
}