Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Sarvam AI / Saaras

Saaras v3

Speech recognition with transcription, translation, verbatim, transliteration and code-mixed output modes.

Transcription
availableClosed weightsSpeech recognition
Official documentation (opens in a new tab)Reviewed 13 Oct 2026
7 of 18 specification fields collected
Audio durationREST up to 30 seconds; batch up to 2 hours per fileSee specification notes below
Audio formatsNot collectedAwaiting source review
Input billingSee ratesSarvam API
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)
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)
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
View all 23 entries
  • Bengali
  • English
  • Gujarati
  • Hindi
  • Kannada
  • Malayalam
  • Marathi
  • Odia
  • Punjabi
  • Tamil
  • Telugu
  • Assamese
  • Bodo
  • Dogri
  • Konkani
  • Kashmiri
  • Maithili
  • Manipuri
  • Nepali
  • Sanskrit
  • Santali
  • Sindhi
  • Urdu
Source for Languages (opens in a new tab)

Indian English plus 22 Indic languages.

Context caching
Not collected
outputModes
transcribe, translate, verbatim, translit, codemixSource for outputModes (opens in a new tab)

Translation mode targets English; these modes do not establish arbitrary language-to-language speech translation.

Audio duration limits
REST up to 30 seconds; batch up to 2 hours per fileSource for Audio duration limits (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.

Sarvam API

Provider source (opens in a new tab)
API model ID
saaras:v3
Access
Sarvam API subscription key; credit balance and account rate limits apply.
Regions
Model-specific regional routing and contractual data residency require verification.
API endpoint URL
https://api.sarvam.ai/speech-to-textSource for API endpoint URL (opens in a new tab)
Audio duration limits
30 secondsSource for Audio duration limits (opens in a new tab)
Streaming
Not supportedSource for Streaming (opens in a new tab)
Word timestamps
Not supportedSource for Word timestamps (opens in a new tab)

with_timestamps returns phrase/sentence chunks despite a response field named words.

Speaker diarization
Not supportedSource for Speaker diarization (opens in a new tab)

Use Batch for diarization.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
transcriptionSpeech recognition / translation1₹30.00audio hour

Rate conditions

  1. Managed speech service; billed per second. Diarization is priced separately; no batch discount is assumed.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

Sarvam API

Provider source (opens in a new tab)
API model ID
saaras:v3
Access
Sarvam API subscription key; credit balance and account rate limits apply.
Regions
Model-specific regional routing and contractual data residency require verification.
API endpoint URL
https://api.sarvam.ai/speech-to-text/job/v1Source for API endpoint URL (opens in a new tab)
Batch processing
SupportedSource for Batch processing (opens in a new tab)
Speaker diarization
SupportedSource for Speaker diarization (opens in a new tab)

Beta diarization option; a separate higher rate applies.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
transcriptionSpeech recognition / translation1₹30.00audio hour
transcriptionSpeech recognition with diarization2₹45.00audio hour

Rate conditions

  1. Managed speech service; billed per second. Diarization is priced separately; no batch discount is assumed.
  2. Total diarized service rate, not an additional ₹45 surcharge; billed per second.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

Sarvam API

Provider source (opens in a new tab)
API model ID
saaras:v3
Access
Sarvam API subscription key; credit balance and account rate limits apply.
Regions
Model-specific regional routing and contractual data residency require verification.
API endpoint URL
wss://api.sarvam.ai/speech-to-text/wsSource for API endpoint URL (opens in a new tab)
Streaming
SupportedSource for Streaming (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 13 Oct 2026
CategoryUsageRateUnit
transcriptionSpeech recognition / translation1₹30.00audio hour

Rate conditions

  1. Managed speech service; billed per second. Diarization is priced separately; no batch discount is assumed.
Pricing source (opens in a new tab)

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

Amazon SageMaker · Sarvam package

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

A route or service is documented, but a stable version-specific checkpoint selector has not been established.

Access
Subscribe to the Sarvam AWS Marketplace model package and deploy in your AWS account. Package ARN and deployment configuration are customer-specific.
Regions
Customer-selected AWS region, subject to the package listing’s supported regions.
Usage metering
AWS instance usage plus software fee. No numerical Marketplace rate verified; managed per-character/per-hour rates do not apply.Source for Usage metering (opens in a new tab)

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

Provider documents network-isolated deployment inside the customer VPC, with no outbound internet from the container. This applies to the self-hosted package, not the managed API.

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.

Sarvam API

saaras:v3
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

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

Sarvam API

saaras:v3
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

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

Sarvam API

saaras:v3
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

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

Amazon SageMaker · Sarvam package

Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

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

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

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

Inferred for this exact native API release/service from provider documentation and review of Sarvam’s public weight inventory. No weights for this identity were identified. Public Sarvam language/translation weights and hosted third-party weights are excluded. A named service or version does not establish an independently disclosed checkpoint.

Classification evidence (opens in a new tab)

Sources reviewed