Instructions to use wavestonewavelets/CompressedSarvam with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use wavestonewavelets/CompressedSarvam with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="wavestonewavelets/CompressedSarvam", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("wavestonewavelets/CompressedSarvam", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use wavestonewavelets/CompressedSarvam with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "wavestonewavelets/CompressedSarvam"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "wavestonewavelets/CompressedSarvam",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/wavestonewavelets/CompressedSarvam

SGLang

How to use wavestonewavelets/CompressedSarvam with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "wavestonewavelets/CompressedSarvam" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "wavestonewavelets/CompressedSarvam",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "wavestonewavelets/CompressedSarvam" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "wavestonewavelets/CompressedSarvam",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use wavestonewavelets/CompressedSarvam with Docker Model Runner:
```
docker model run hf.co/wavestonewavelets/CompressedSarvam
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

This model, is the first iteration of team WavestoneWavelets for the Sustainable AI Coalition model compression challenge. For now this is the original model, and no compression technique has been used.

Introduction

Sarvam-30B is an advanced Mixture-of-Experts (MoE) model with 2.4B non-embedding active parameters, designed primarily for practical deployment. It combines strong reasoning, reliable coding ability, and best-in-class conversational quality across Indian languages. Sarvam-30B is built to run reliably in resource-constrained environments and can handle multilingual voice calls while performing tool calls.

A major focus during training was the Indian context and languages, resulting in state-of-the-art performance across 22 Indian languages for its model size.

Sarvam-30B is open-sourced under the Apache License. For more details, see our blog.

Compression strategy

For now, the compression is only a 8 bits quantization

Citation

@misc{sarvam_sovereign_models,
  title        = {Introducing Sarvam's Sovereign Models},
  author       = {{Sarvam Foundation Models Team}},
  year         = {2026},
  howpublished = {\url{https://www.sarvam.ai/blogs/sarvam-30b-105b}},
  note         = {Accessed: 2026-03-03}
}

Downloads last month: 19

Safetensors

Model size

32B params

Tensor type

F32