Instructions to use sarvamai/sarvam-m with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use sarvamai/sarvam-m with Transformers:

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

pipe = pipeline("text-generation", model="sarvamai/sarvam-m")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("sarvamai/sarvam-m")
model = AutoModelForCausalLM.from_pretrained("sarvamai/sarvam-m", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use sarvamai/sarvam-m with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/sarvamai/sarvam-m

SGLang

How to use sarvamai/sarvam-m 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 "sarvamai/sarvam-m" \
    --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": "sarvamai/sarvam-m",
		"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 "sarvamai/sarvam-m" \
        --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": "sarvamai/sarvam-m",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use sarvamai/sarvam-m with Docker Model Runner:
```
docker model run hf.co/sarvamai/sarvam-m
```

mohit-sarvam commited on May 23, 2025

Commit

2bf88b5

verified ·

1 Parent(s): a198288

Update README.md

Browse files

Files changed (1) hide show

README.md +8 -5

README.md CHANGED Viewed

@@ -116,7 +116,7 @@ messages = [
 response1 = client.chat.completions.create(
     model=model_name,
     messages=messages,
-    reasoning_effort="medium",  # Optional reasoning mode
     max_completion_tokens=4096,
 )
 print("First response:", response1.choices[0].message.content)
@@ -135,7 +135,7 @@ messages.extend(
 response2 = client.chat.completions.create(
     model=model_name,
     messages=messages,
-    reasoning_effort="high",
     max_completion_tokens=8192,
 )
 print("Follow-up response:", response2.choices[0].message.content)
@@ -143,6 +143,8 @@ print("Follow-up response:", response2.choices[0].message.content)
 Refer to API docs here: [sarvam API docs](https://docs.sarvam.ai/api-reference-docs/introduction)
 # VLLM Deployment
 For easy deployment, we can use `vllm>=0.8.5` and create an OpenAI-compatible API endpoint with `vllm serve sarvamai/sarvam-m`
@@ -195,7 +197,8 @@ The repo contains bf16 and q8 gguf files built using https://github.com/ggml-org
 You can use the model using cli as explained in docs https://github.com/ggml-org/llama.cpp/tree/master/tools/main
 Example Command:
-`./build/bin/llama-cli -i -m /projects/data/romit_sarvam_ai/models/gguf/sarvam-m-q8_0.gguf -c 8192 -t 16`
-We got about 4 tokens per second on 16 cores.

 response1 = client.chat.completions.create(
     model=model_name,
     messages=messages,
+    reasoning_effort="medium",  # Enable thinking mode
     max_completion_tokens=4096,
 )
 print("First response:", response1.choices[0].message.content)
 response2 = client.chat.completions.create(
     model=model_name,
     messages=messages,
+    reasoning_effort="medium",
     max_completion_tokens=8192,
 )
 print("Follow-up response:", response2.choices[0].message.content)
 Refer to API docs here: [sarvam API docs](https://docs.sarvam.ai/api-reference-docs/introduction)
+The model has only two modes: `think` and `no-think`. If `reasoning_effort` is set then thinking mode is on otherwise it is off.
 # VLLM Deployment
 For easy deployment, we can use `vllm>=0.8.5` and create an OpenAI-compatible API endpoint with `vllm serve sarvamai/sarvam-m`
 You can use the model using cli as explained in docs https://github.com/ggml-org/llama.cpp/tree/master/tools/main
 Example Command:
+```
+./build/bin/llama-cli -i -m /projects/data/romit_sarvam_ai/models/gguf/sarvam-m-q8_0.gguf -c 8192 -t 16
+```
+We got about 4 tokens per second on 16 cores with q8 model.