Indic LM Platform

In plain words

sarvam-translate is a multimodal translation and assistant model from Sarvam AI, strictly optimised for enterprise Indic workflows.

How to use it

Before you start

Prerequisites required to run the model locally

To run sarvam-translate model locally on your computer, the following requirements must be met:

Install and set up Ollama.
Ensure your computer has enough RAM to run the model you plan to download.

For 3B models: Minimum 8 GB of RAM.
For 7B–8B models: Minimum 16 GB of RAM.
For 13B–32B models: Minimum 32 GB of RAM.

Quantised model: It is recommended to use 4-bit Quantized, K-measurement, Small model format. Easy to download and to run your system.
Storage: 5 to 20 GB of free SSD space to store downloaded model files.
Interface: Terminal or command line utility to run basic startup prompts.

No technical setup needed — anyone can try sarvam-translate using below mentioned steps.

Downloading the sarvam-translate GGUF file

Go to the sarvam-translate GGUF file on Hugging Face. Click the copy icon next to the repository name to copy the model path. (See image below.)

Run the model in your terminal

Open a terminal window and run the following command and paste the copied GGUF repository as mentioned below:

ollama run hf.co/PASTE_THE_COPIED_HERE

Press Enter.

Note: After pasting the repository, your command will look like this:

ollama run hf.co/trysem/sarvam-translate-Q4_K_M-GGUF

Wait for the model to load

Ollama will download and initialise the model. Once ready, you'll see the following prompt:

>>> Send a message (/? for help)

Start chatting

Type your message at the prompt and press Enter to interact with the model.

Exit the chat

To end the session, type:

/bye

Then press Enter.

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 dataFine-tuned on a mix of cleaned historical parallel corpora (including AI4Bharat's BPCC) and newly generated translation pairs spanning scientific, conversational, and structurally complex formats such as code and LaTeX.

SizeFine-tuned from Google's Gemma-3-4B (full fine-tune plus LoRA) in partnership with AI4Bharat; supports an 8,192-token context window for document-level translation.

LicenceReleased under GPL-3.0 — a copyleft licence requiring derivative works to also be distributed under GPL-3.0, stricter than the Apache 2.0 used for Sarvam's other models.

Commercial cost₹20 per 10,000 characters via API; free to self-host (GPL-3.0)

Version history1.0 · June 2025 — Initial release, built on Gemma-3-4B, developed with AI4Bharat

About the maker

Who builds and maintains sarvam-translate.

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 →