Indic LM Platform

In plain words

Sarvam 30B is a highly capable, 30-billion-parameter reasoning model from Sarvam AI, optimised for Indian linguistic tasks and code generation.

How to use it

🚧

This section is coming soon.

A step-by-step guide for Sarvam 30B 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 dataTrained on ~16 trillion tokens (code, web, maths, multilingual data with synthetic augmentation), with significant allocation to 10 major Indian languages; post-trained via SFT plus GRPO-based reinforcement learning, mirroring Sarvam 105B's pipeline.

SizeMixture-of-experts model (~30-32B total / 2.4B active parameters), 19 layers with 128 experts plus one shared expert and grouped-query attention, optimised for low-latency inference.

LicenceOpen-sourced under the Apache License 2.0, weights downloadable from Hugging Face and AI Kosh, free for commercial and non-commercial use.

Commercial cost₹4/1M input tokens · ₹16/1M output tokens via API; free to self-host (Apache 2.0)

Version history1.0 · March 2026 — Unveiled at the India-AI Impact Summit, open-sourced under Apache 2.0

About the maker

Who builds and maintains Sarvam 30B.

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 →