Sarvam AI

Sarvam AI is India's full-stack sovereign AI platform building large language models, speech, and translation systems for Indian languages. The Sarvam API serves chat completions (Sarvam-M / Sarvam-30B / Sarvam-105B), speech-to-text (Saaras / Saarika), text-to-speech (Bulbul), translation (Mayura / Sarvam-Translate), transliteration, and language identification across 10-22 Indic languages via a REST interface using an api-subscription-key header.

7 APIs 0 Features
AILLMSpeech to TextText to SpeechTranslationIndian Languages

APIs

Sarvam AI Chat Completions (LLM) API

OpenAI-style chat completions over Sarvam's Indic LLMs (Sarvam-M, Sarvam-30B, Sarvam-105B) with streaming, tool use, reasoning effort, and wiki-grounding for Indian-language rea...

Sarvam AI Speech-to-Text API

Multipart audio transcription for Indian languages using the Saaras and Saarika speech-recognition models, with auto language detection, word timestamps, and speaker diarization.

Sarvam AI Speech-to-Text Translate API

Transcribes Indic-language audio and translates the result into English in a single multipart request using the Saaras family of models.

Sarvam AI Text-to-Speech API

Natural-voice speech synthesis for Indian languages using the Bulbul models (v2 and v3) with 30+ speakers, configurable pace, pitch, sample rate, and multiple output audio codec...

Sarvam AI Translate API

Text translation across Indian languages using Mayura (12 languages, multiple tones) and Sarvam-Translate (all 22 scheduled languages), with tone modes, output-script control, a...

Sarvam AI Transliterate API

Converts text between scripts while preserving the same language, with spoken-form conversion and international or native numeral formatting.

Sarvam AI Language Identification API

Automatically detects the language and script of input text, returning BCP-47 language and script codes for routing to the appropriate Indic-language model.

Collections

Pricing Plans

Sarvam Ai Plans Pricing

3 plans

PLANS

Rate Limits

Sarvam Ai Rate Limits

4 limits

RATE LIMITS

FinOps

Event Specifications

Sarvam AI Chat Completions Streaming (HTTP + SSE)

AsyncAPI 2.6 description of Sarvam AI's **chat completion streaming** surface. Sarvam's chat completions endpoint streams partial results as **HTTP Server-Sent Events (SSE)** ov...

ASYNCAPI

Resources

👥
GitHubOrganization
GitHubOrganization
🔗
LinkedIn
LinkedIn
🔗
Website
Website
🔗
Documentation
Documentation
🔗
Plans
Plans
🔗
RateLimits
RateLimits
🔗
FinOps
FinOps

Sources

Raw ↑
opencollection: 1.0.0
info:
  name: Sarvam AI API
  version: '1.0'
request:
  auth:
    type: apikey
    in: header
    name: api-subscription-key
    value: '{{apiSubscriptionKey}}'
items:
- info:
    name: Chat
    type: folder
  items:
  - info:
      name: Create a chat completion
      type: http
    http:
      method: POST
      url: https://api.sarvam.ai/v1/chat/completions
      body:
        type: json
        data: "{\n  \"model\": \"sarvam-m\",\n  \"messages\": [ { \"role\": \"user\", \"content\": \"Namaste!\" } ],\n  \"\
          stream\": false\n}"
    docs: Generates a model response using Sarvam's Indic LLMs (sarvam-m, sarvam-30b, sarvam-105b). Set stream=true for SSE
      streaming.
- info:
    name: Speech to Text
    type: folder
  items:
  - info:
      name: Transcribe audio to text
      type: http
    http:
      method: POST
      url: https://api.sarvam.ai/speech-to-text
      body:
        type: multipart-form
        data: []
    docs: Transcribes an audio file in an Indian language using the Saaras or Saarika models.
  - info:
      name: Transcribe and translate audio to English
      type: http
    http:
      method: POST
      url: https://api.sarvam.ai/speech-to-text-translate
      body:
        type: multipart-form
        data: []
    docs: Transcribes Indic-language audio and translates the result into English using the Saaras family of models.
- info:
    name: Text to Speech
    type: folder
  items:
  - info:
      name: Convert text to speech
      type: http
    http:
      method: POST
      url: https://api.sarvam.ai/text-to-speech
      body:
        type: json
        data: "{\n  \"text\": \"नमस्ते\",\n  \"target_language_code\": \"hi-IN\",\n  \"model\": \"bulbul:v3\",\n  \"speaker\"\
          : \"shubh\"\n}"
    docs: Synthesizes natural speech for Indian languages using the Bulbul models. Returns base64-encoded audio.
- info:
    name: Translate
    type: folder
  items:
  - info:
      name: Translate text between Indian languages
      type: http
    http:
      method: POST
      url: https://api.sarvam.ai/translate
      body:
        type: json
        data: "{\n  \"input\": \"How are you?\",\n  \"source_language_code\": \"en-IN\",\n  \"target_language_code\": \"hi-IN\"\
          ,\n  \"model\": \"mayura:v1\"\n}"
    docs: Translates text across Indian languages using Mayura or Sarvam-Translate.
- info:
    name: Transliterate
    type: folder
  items:
  - info:
      name: Transliterate text between scripts
      type: http
    http:
      method: POST
      url: https://api.sarvam.ai/transliterate
      body:
        type: json
        data: "{\n  \"input\": \"namaste\",\n  \"source_language_code\": \"en-IN\",\n  \"target_language_code\": \"hi-IN\"\
          \n}"
    docs: Converts text between scripts while preserving the same language.
- info:
    name: Language Identification
    type: folder
  items:
  - info:
      name: Identify the language of text
      type: http
    http:
      method: POST
      url: https://api.sarvam.ai/text-lid
      body:
        type: json
        data: "{\n  \"input\": \"आप कैसे हैं?\"\n}"
    docs: Detects the language and script of input text, returning BCP-47 language and script codes.