How to get repeatable results using seed

The seed parameter lets you make model outputs deterministic — meaning you’ll get the same output every time for the same prompt and settings.

Great for testing, debugging, or consistent experiences in production.

What it does:

• When you pass a seed, the model uses that value to initialize its random number generator.
• This allows it to sample deterministically — so the same prompt and settings will always produce the same result.
• If you don’t pass a seed, the output will naturally vary on each request.
• Note: This feature is currently in Beta — the system will make a best effort to sample deterministically when seed is used.

Format and Range:

👉 Typically, values like 42, 1234, or any stable number will work fine.

When to use seed:

First, install the SDK:

$
pip install -Uqq sarvamai

Then use the following Python code:

1
from sarvamai import SarvamAI
2
3
# Initialize the SarvamAI client with your API key
4
client = SarvamAI(api_subscription_key="YOUR_SARVAM_API_KEY")
5
6
# Example: Using `seed` to repeat responses
7
response = client.chat.completions(
8
    model="sarvam-105b",
9
    messages=[
10
        {"role": "system", "content": "You are a poetry assistant."},
11
        {"role": "user", "content": "Write a short poem about the moon."}
12
    ],
13
    seed=12345  # Ensures the same poem is returned every time (best effort)
14
)
15
16
# Receive assistant's reply as output
17
print(response.choices[0].message.content)