generated: '2026-07-21'
method: derived
source: openapi/sarvam-openapi-original.json
notes: >-
  Derived from OpenAPI paths and identifier fields. Sarvam is largely a
  stateless inference API (transcribe/translate/synthesize/chat return results
  inline); the only persistent, addressable resources are async Jobs and
  Pronunciation Dictionaries.
entities:
- name: Job
  id_field: job_id
  kinds: [speech-to-text, speech-to-text-translate, doc-digitization]
  lifecycle: [initialise, upload-files, start, status, download-files]
  operations:
  - openapi/sarvam-openapi-original.json#initialise
  - openapi/sarvam-openapi-original.json#get-upload-links
  - openapi/sarvam-openapi-original.json#start
  - openapi/sarvam-openapi-original.json#get-status
  - openapi/sarvam-openapi-original.json#get-download-links
- name: PronunciationDictionary
  id_field: dict_id
  operations:
  - openapi/sarvam-openapi-original.json#create
  - openapi/sarvam-openapi-original.json#list
  - openapi/sarvam-openapi-original.json#update
  - openapi/sarvam-openapi-original.json#delete
  - openapi/sarvam-openapi-original.json#get
- name: ChatCompletion
  id_field: null
  stateless: true
  operations:
  - openapi/sarvam-openapi-original.json#completions
relationships:
- from: Job
  to: PronunciationDictionary
  type: none
  notes: Independent resources; no cross-references between the two entity families.