Indic LM Platform

In plain words

Sarvam-M is a 24B chat-tuned hybrid reasoning model from Sarvam AI, covering 11 Indian languages and targeted at robust, complex linguistic tasks.

How to use it

🚧

This section is coming soon.

A step-by-step guide for Sarvam-M will appear here soon.

Languages & scripts supported

Works in these languages, in both native script and Roman typing.

हिन्दी Hindiதமிழ் Tamilతెలుగు Teluguമലയാളം Malayalamਪੰਜਾਬੀ Punjabiଓଡ଼ିଆ Odiaગુજરાતી Gujaratiमराठी Marathiಕನ್ನಡ Kannadaবাংলা Bengali+ Roman & code-mixed

Strengths & limits

An honest look at what it does well and where it struggles.

Good at

✓Native language processing

✓Text generation

Where it struggles

!Highly complex code generation

!Low-resource dialects without fine-tuning

Technical details & licence

Training dataPost-trained on top of Mistral-Small-3.1-24B via a three-stage pipeline: SFT on 3.7M curated prompts (heavily Hindi and other Indic languages), RLVR (GRPO) on maths, coding, and translation tasks, then FP8 quantisation for inference efficiency.

SizeInstruction/reasoning-tuned derivative of Mistral-Small-3.1-24B-Base, a dense transformer with a hybrid "think"/"non-think" reasoning mode and a 32,768-token context window.

LicenceReleased under Apache 2.0, the same licence as its Mistral-Small base, allowing free commercial use, modification, and self-hosting.

Commercial costFree to self-host (Apache 2.0); also available via third-party inference providers

Version history1.0 · May 2025 — Initial release — hybrid reasoning model built on Mistral-Small-3.1-24B

About the maker

Who builds and maintains Sarvam-M.

Sarvam AI

5 models

Sarvam AI is an Indian artificial-intelligence company building voice-first, multilingual AI made for India. Founded in 2023 by researchers behind several well-known open Indic models, it focuses on practical, affordable models that work naturally across many Indian languages.

View their models →See all organisations →