SambaNova Integration¶

Instructor supports SambaNova's LLM API, allowing you to use structured outputs with their models.

Installation¶

pip install "instructor[openai]"

Basic Usage¶

import instructor
import os
from pydantic import BaseModel

client = instructor.from_provider("sambanova/Meta-Llama-3.1-405B-Instruct")

class User(BaseModel):
    name: str
    age: int

user = client.chat.completions.create(
    messages=[
        {"role": "user", "content": "Ivan is 28"},
    ],
    response_model=User,
)

print(user)
# > User(name='Ivan', age=28)

Async Usage¶

import instructor
import os
from pydantic import BaseModel

client = instructor.from_provider(
    "sambanova/Meta-Llama-3.1-405B-Instruct",
    async_client=True,
)

class User(BaseModel):
    name: str
    age: int

async def get_user():
    user = await client.chat.completions.create(
        messages=[
            {"role": "user", "content": "Ivan is 28"},
        ],
        response_model=User,
    )
    return user

# Run with asyncio
import asyncio
user = asyncio.run(get_user())
print(user)
# > User(name='Ivan', age=28)

Available Models¶

Check the SambaNova documentation for the latest model offerings and capabilities.