SambaNova Python SDK

1.8.2 verified Sat May 09 auth: no python

The official Python library for the SambaNova API. Provides synchronous and asynchronous clients for chat completions, embeddings, audio, responses, and vision. Actively maintained, version 1.8.2, with frequent releases.

pip install sambanova

Common errors

error sambanova.lib.SambaNovaError: api_key is required ↓

cause No API key provided.

fix

Set the SAMBANOVA_API_KEY environment variable or pass api_key='your_key' to SambaNovaClient().

error ModuleNotFoundError: No module named 'sambanova' ↓

cause Library not installed.

fix

Run 'pip install sambanova'.

error AttributeError: module 'sambanova' has no attribute 'Client' ↓

cause Wrong import: tried 'from sambanova import Client'.

fix

Use 'from sambanova import SambaNovaClient'.

Warnings

breaking API key must be set either via environment variable SAMBANOVA_API_KEY or passed directly. The library will raise an authentication error if missing. ↓

fix Set os.environ['SAMBANOVA_API_KEY'] or pass api_key argument to SambaNovaClient().

gotcha The model name must match SambaNova's model IDs exactly (e.g., 'Meta-Llama-3.1-8B-Instruct', 'Llama-3.2-11B-Vision-Instruct'). Using a generic OpenAI model name will fail. ↓

fix Refer to SambaNova documentation for exact model IDs.

deprecated Support for Python 3.8 was dropped after version 1.5.0. The library now requires Python >=3.9. ↓

fix Upgrade Python to 3.9 or higher.

Imports

SambaNovaClient
wrong
```
from sambanova import Client
```
correct
```
from sambanova import SambaNovaClient
```
Client is not exported; must use SambaNovaClient.
AsyncSambaNovaClient
```
from sambanova import AsyncSambaNovaClient
```
For async usage; also available via sambanova.AsyncSambaNovaClient.

Quickstart

Initialize the sync client with an API key (env var recommended) and call chat completions.

from sambanova import SambaNovaClient

client = SambaNovaClient(
    api_key=os.environ.get('SAMBANOVA_API_KEY', ''),
)
try:
    response = client.chat.completions.create(
        model='Meta-Llama-3.1-8B-Instruct',
        messages=[{'role': 'user', 'content': 'Hello'}],
    )
    print(response.choices[0].message.content)
except Exception as e:
    print(f'Error: {e}')