> For clean Markdown of any page, append `.md` to the page URL.
> For a complete documentation index, see https://docs.sarvam.ai/llms.txt.
> For full documentation content in one file, see https://docs.sarvam.ai/llms-full.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.sarvam.ai/_mcp/server.

# Models

> Complete overview of Sarvam AI's specialized models for Indian languages. Choose the right model for your use case - from speech processing to text generation, translation, and document intelligence.

Sarvam AI provides a purpose-built AI stack for building applications in Indian languages. Our models span speech-to-text, speech translation, text translation, and high-quality text-to-speech, designed specifically for India's linguistic diversity, accents, and real-world usage patterns.

Each model is trained and evaluated on Indian languages and culturally grounded data, enabling higher accuracy in production scenarios. With simple, well-documented APIs and predictable performance, developers can build, deploy, and scale India-first AI experiences without managing model complexity.

> **Tip**
>
> New to building for Indian languages? Start with [Building for Indian Languages](/api/getting-started/building-for-india), a practical guide to language coverage, code-mixing, scripts, native numerals, 8kHz telephony audio, and pronunciation control.

## Model Selection Guide

**Sarvam Models**: trained for Indian languages:

| Model                                                                    | API                   | Description                                                                                                                                                                                                        |
| ------------------------------------------------------------------------ | --------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| [Saaras v4](/api/getting-started/models/saaras)                          | Speech to Text        | State-of-the-art ASR with 23 language support (22 Indian + Global/Indian English) and multiple output modes: transcribe, translate, verbatim, translit, codemix. Default, recommended model; v3 remains available. |
| [Bulbul v3](/api/getting-started/models/bulbul)                          | Text to Speech        | Natural-sounding voices for 11 languages (10 Indian + English) with customizable pitch, pace, and speaker options.                                                                                                 |
| [Sarvam Voice Cloning](/api/api-guides-tutorials/voice-cloning/overview) | Voice Cloning         | Create a voice from a short reference clip and synthesize speech with it across 12 Indian languages - clone inline per request, or save the voice once and reuse its ID.                                           |
| [Sarvam-105B](/api/getting-started/models/sarvam-105b)                   | Chat Completion       | 105B parameter flagship model, `sarvam-105b` for complex reasoning and agentic tasks, plus `sarvam-105b-conversations` for real-time dialogue and voice agents.                                                    |
| [Mayura](/api/getting-started/models/mayura)                             | Text Translation      | High-quality translation between 11 languages (10 Indian + English) with context preservation.                                                                                                                     |
| [Sarvam-Translate](/api/getting-started/models/sarvam-translate)         | Text Translation      | Extended translation support for all 23 languages (22 Indian + English) with superior accuracy.                                                                                                                    |
| [Sarvam Vision](/api/getting-started/models/sarvam-vision)               | Document Intelligence | Extract and digitize content from documents in 23 languages with accurate OCR and structured output.                                                                                                               |

## Open-Weight Models

Open-weight models are available on **`/v2`** using the same Sarvam API key and credits.

> **Note**
>
> GLM-5.3, Gemma 4 31B, and DeepSeek V4 Flash are **available in beta** and rolling out
> gradually. See [Access to Beta APIs](/api-reference/beta-apis).

| Model                                                                         |   Context window | Modality            | Reasoning |
| ----------------------------------------------------------------------------- | ---------------: | ------------------- | :-------: |
| [GLM-5.3](/api/getting-started/models/openweight/glm-5-3)                     | 1,048,576 tokens | Text → text         |    Yes    |
| [Gemma 4 31B](/api/getting-started/models/openweight/gemma-4-31b)             |   131,072 tokens | Text + image → text |     No    |
| [DeepSeek V4 Flash](/api/getting-started/models/openweight/deepseek-v4-flash) | 1,048,576 tokens | Text → text         |    Yes    |

[Read more about the open-weight models →](/api/getting-started/models/openweight)

## Language Support Overview

Language coverage varies by model. Check the table below before picking one. Full per-model tables are linked from each model's own page.

| Model                                                                                   | Languages                                                                                                | Status      |
| --------------------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------- | ----------- |
| **Saaras v4** (Speech to Text)                                                          | 23 (22 Indian + Global/Indian English), [full list](/api/getting-started/models/saaras#language-support) | Recommended |
| **Sarvam Translate** (Text Translation)                                                 | 23 (22 Indian + English), [full list](/api/getting-started/models/sarvam-translate#language-support)     | Active      |
| **Sarvam Vision** (Document Intelligence)                                               | 23 (22 Indian + English), [full list](/api/getting-started/models/sarvam-vision#supported-languages)     | Active      |
| **Bulbul v3** (Text to Speech)                                                          | 11 (10 Indian + English), [full list](/api/getting-started/models/bulbul#language-support)               | Active      |
| **Sarvam Voice Cloning**                                                                | 12 Indian languages, [full list](/api/api-guides-tutorials/voice-cloning/supported-languages)            | Active      |
| **Mayura** (Text Translation)                                                           | 11 (10 Indian + English), [full list](/api/getting-started/models/mayura#language-support)               | Active      |
| **Sarvam-105B** (Chat LLM)                                                              | 11 (10 Indian + English), `sarvam-105b`, `sarvam-105b-conversations`                                     | Active      |
| **GLM-5.3** (Chat LLM, [open-weight](/api/getting-started/models/openweight))           | See the model page for language coverage                                                                 | Beta        |
| **Gemma 4 31B** (Chat LLM, [open-weight](/api/getting-started/models/openweight))       | See the model page for language coverage                                                                 | Beta        |
| **DeepSeek V4 Flash** (Chat LLM, [open-weight](/api/getting-started/models/openweight)) | See the model page for language coverage                                                                 | Beta        |

### 23-language set (Saaras v4, Sarvam Translate, Sarvam Vision)

| Language  | Code    |   | Language | Code     |
| --------- | ------- | - | -------- | -------- |
| Hindi     | `hi-IN` |   | Assamese | `as-IN`  |
| Bengali   | `bn-IN` |   | Urdu     | `ur-IN`  |
| Kannada   | `kn-IN` |   | Nepali   | `ne-IN`  |
| Malayalam | `ml-IN` |   | Konkani  | `kok-IN` |
| Marathi   | `mr-IN` |   | Kashmiri | `ks-IN`  |
| Odia      | `od-IN` |   | Sindhi   | `sd-IN`  |
| Punjabi   | `pa-IN` |   | Sanskrit | `sa-IN`  |
| Tamil     | `ta-IN` |   | Santali  | `sat-IN` |
| Telugu    | `te-IN` |   | Manipuri | `mni-IN` |
| English   | `en-IN` |   | Bodo     | `brx-IN` |
| Gujarati  | `gu-IN` |   | Maithili | `mai-IN` |
|           |         |   | Dogri    | `doi-IN` |

### 11-language set (Bulbul v3, Mayura, Sarvam-105B)

| Language | Code    |   | Language  | Code    |
| -------- | ------- | - | --------- | ------- |
| Hindi    | `hi-IN` |   | Kannada   | `kn-IN` |
| Bengali  | `bn-IN` |   | Malayalam | `ml-IN` |
| Tamil    | `ta-IN` |   | Marathi   | `mr-IN` |
| Telugu   | `te-IN` |   | Punjabi   | `pa-IN` |
| Gujarati | `gu-IN` |   | Odia      | `od-IN` |
| English  | `en-IN` |   |           |         |

---

## Use Cases

#### Voice Assistant

### Build a multilingual voice assistant

1. **Speech Input**: Use Saaras v4 with `mode="transcribe"` to convert user speech to text
2. **Understanding**: Process with Sarvam-105B for intelligent responses
3. **Speech Output**: Convert responses to speech with Bulbul

Perfect for customer service, smart home devices, and accessibility applications.

[Learn how to build a voice agent with LiveKit →](/api/integration/build-voice-agent-with-live-kit)

#### Content Localization

### Localize content across Indian languages

Build end-to-end multilingual experiences from audio to translated speech:

1. **Transcribe speech**: Use Saaras v4 with `mode="transcribe"` to accurately convert audio in Indian languages into text.
2. **Translate at scale**: Convert transcripts across 23 languages (22 Indian + English) with Sarvam Translate.
3. **Generate localized audio**: Turn translated text into natural, production-ready speech with Bulbul's high-quality voices.

This gives you a complete pipeline, speech → text → translation → localized audio, enabling developers to deliver fully localized content for apps, videos, learning platforms, and product experiences with minimal engineering effort.

#### Call Center Analytics

### Analyze multilingual customer interactions

1. **Speech Recognition**: Convert calls to text with Saaras v4 (`mode="transcribe"`)
2. **Translation**: Use Saaras v4 with `mode="translate"` for English output
3. **Insights**: Extract patterns and sentiment from conversations

Essential for customer experience optimization and compliance monitoring.

[Learn how to analyze calls with our analytics cookbook →](/api/cookbook/guides/call-analytics-pipeline)

#### Educational Platform

### Create inclusive learning experiences

1. **Content Translation**: Make materials accessible in 23 languages (22 Indian + English)
2. **Audio Learning**: Generate pronunciation guides with Bulbul
3. **Interactive Chat**: Enable Q\&A with Sarvam-105B in native languages

Perfect for online education, language learning, and skill development platforms.

#### Document Processing

### Extract and digitize document content

1. **Document Upload**: Submit PDFs or images of documents in Indian languages
2. **Text Extraction**: Use Sarvam Vision for accurate OCR across 23 languages
3. **Structured Output**: Get clean HTML, Markdown, or JSON output with tables preserved

Ideal for digitizing government forms, invoices, legal documents, and historical records in Indian languages.

[Learn more about Document AI →](/api/api-guides-tutorials/document-intelligence/overview)

#### Legacy Models

The following models are still available but are being phased out. We recommend migrating to the newer models listed above.

#### [Sarvam-M: Chat & Reasoning (Deprecated)](/api/getting-started/models/sarvam-m)

24B parameter multilingual chat model with hybrid reasoning. Deprecated and no longer available through the API. Migrate to Sarvam-105B.

#### [Sarvam-30B: Chat LLM (Deprecated)](/api/getting-started/models/sarvam-30b)

30B parameter multilingual chat model with strong reasoning and conversational capabilities. Deprecated. Migrate to Sarvam-105B.

> Complete overview of Sarvam AI's specialized models for Indian languages. Choose the right model for your use case - from speech processing to text generation, translation, and document intelligence.

## Docs

- [Saaras](https://docs.sarvam.ai/api/getting-started/models/saaras.md): Saaras v3 and v4 - Domain-aware speech translation models that convert speech directly to English text with enhanced telephony support and intelligent entity preservation.
- [Bulbul](https://docs.sarvam.ai/api/getting-started/models/bulbul.md): Bulbul v3 - High-quality multilingual text-to-speech model for Indian languages with natural prosody and 30+ speaker voices.
- [Sarvam-105B](https://docs.sarvam.ai/api/getting-started/models/sarvam-105b.md): Sarvam-105B - 105B parameter flagship multilingual language model delivering state-of-the-art performance on Indian language understanding, reasoning, and generation tasks.
- [Open-Weight Models](https://docs.sarvam.ai/api/getting-started/models/openweight.md): Open-weight models served on Sarvam infrastructure through the OpenAI-compatible chat completions API, using your existing Sarvam API key and credits.
- [GLM-5.3](https://docs.sarvam.ai/api/getting-started/models/openweight/glm-5-3.md): GLM-5.3 is a large-scale reasoning model built for complex software engineering and long-horizon agent tasks. It supports text input and output with a 1,048,576-token context window.
- [Gemma 4 31B](https://docs.sarvam.ai/api/getting-started/models/openweight/gemma-4-31b.md): Gemma 4 31B is a 31B-parameter instruction-tuned model for text and image understanding. It supports a 131,072-token context window and tool calling.
- [DeepSeek V4 Flash](https://docs.sarvam.ai/api/getting-started/models/openweight/deepseek-v4-flash.md): DeepSeek V4 Flash is a general-purpose open-weight reasoning model with text input and output, a 1,048,576-token context window, and tool calling.
- [Mayura](https://docs.sarvam.ai/api/getting-started/models/mayura.md): Mayura - Advanced multilingual translation model for Indian languages with customizable translation styles, script control, and intelligent code-mixed content handling.
- [Sarvam Translate](https://docs.sarvam.ai/api/getting-started/models/sarvam-translate.md): Sarvam Translate - Comprehensive translation model supporting all 22 official Indian languages with formal translation style and structured text optimization.
- [Sarvam Vision](https://docs.sarvam.ai/api/getting-started/models/sarvam-vision.md): Sarvam Vision - A 3B parameter multimodal model delivering world-class Document Intelligence and visual understanding with unmatched accuracy for 23 languages (22 Indian + English).
- [Sarvam-30B (Deprecated)](https://docs.sarvam.ai/api/getting-started/models/sarvam-30b.md): Sarvam-30B - 30B parameter multilingual language model optimized for Indian languages with strong reasoning, coding, and conversational capabilities. Deprecated; migrate to Sarvam-105B.
- [Sarvam-M (Deprecated)](https://docs.sarvam.ai/api/getting-started/models/sarvam-m.md): Sarvam-M - deprecated 24B parameter multilingual, hybrid-reasoning language model with strong Indian language benchmark performance.
- [Saarika](https://docs.sarvam.ai/api/getting-started/models/saarika.md): Saarika v2.5 - High-accuracy speech recognition model for Indian languages with superior multi-speaker handling, telephony optimization, and automatic code-mixing support.