SambaNova LangChain Integration

1.1.0 · active · verified Thu Apr 16

An integration package connecting SambaNova and LangChain, enabling developers to leverage SambaNova's large language models. It provides chat models (`ChatSambaNova`) and embedding models (`SambaNovaEmbeddings`) for seamless interaction with SambaCloud and SambaStack. The current version is 1.1.0, with a regular release cadence that incorporates new features and ensures compatibility with LangChain updates.

Common errors

```
ModuleNotFoundError: No module named 'langchain_sambanova'
```
cause The `langchain-sambanova` package is not installed in the current Python environment.

fix Run `pip install langchain-sambanova` to install the package. Ensure your environment is activated if using virtual environments.
```
ValueError: Did not find sensibility_api_key, please add an environment variable SAMBANOVA_API_KEY
```
cause The `SAMBANOVA_API_KEY` environment variable, which is required for authenticating with SambaNova services, has not been set.

fix Set the environment variable: `export SAMBANOVA_API_KEY="your_api_key_here"` (replace with your actual API key from cloud.sambanova.ai). For SambaStack, also set `SAMBANOVA_API_BASE`.
```
Failed to serialize object: The `content_blocks` field for `AIMessage` is not supported with this `langchain-core` version.
```
cause An older version of `langchain-core` or `langchain-sambanova` is being used which does not correctly handle the `AIMessage` content block format, a common issue fixed in `langchain-sambanova` v1.1.0.

fix Upgrade both `langchain-sambanova` and `langchain-core` to their latest versions: `pip install -U langchain-sambanova langchain-core`.

Warnings

breaking Version 1.1.0 includes fixes related to 'langchain-core 1.2+ security restrictions' and handling of 'AIMessage content_blocks format'. Older code relying on previous serialization behavior or specific AIMessage structures with `langchain-core < 1.2` might break or behave unexpectedly.
Fix: Upgrade to `langchain-sambanova>=1.1.0` and `langchain-core>=1.2.0`. Review any custom serialization logic or message parsing that interacts with `AIMessage` content.
gotcha Authentication requires setting environment variables `SAMBANOVA_API_KEY` and, for SambaStack users, `SAMBANOVA_API_BASE`. Failing to set these will result in authentication errors.
Fix: Ensure `export SAMBANOVA_API_KEY="your-key-here"` is run in your environment. For SambaStack, also set `export SAMBANOVA_API_BASE="your-base-url-here"`.
breaking The jump from `v0.2.0` to `v1.0.0` often indicates significant changes and potential API breakage, even if not explicitly detailed in the changelog as 'breaking'. Users upgrading across this major version boundary should expect refactorings.
Fix: Refer to the official SambaNova LangChain documentation and the library's GitHub README for any migration guides or updated API signatures when upgrading to version 1.0.0 or higher.

Install

pip install langchain-sambanova Install latest version

Imports

ChatSambaNova
```
from langchain_sambanova import ChatSambaNova
```
Primary class for interacting with SambaNova chat models.
SambaNovaEmbeddings
```
from langchain_sambanova import SambaNovaEmbeddings
```
Primary class for interacting with SambaNova embedding models.
SambaStudio
wrong:
```
from langchain_sambanova import SambaStudio
```
correct:
```
from langchain_community.llms import SambaStudio
```
While 'SambaStudio' exists in `langchain_community`, the dedicated `langchain-sambanova` package provides `ChatSambaNova` and `SambaNovaEmbeddings` as the current and recommended interfaces for SambaNova models. The `langchain_community.llms.SambaStudio` is part of an older or broader integration.

Quickstart

This quickstart demonstrates how to set up and use the `ChatSambaNova` model for chat completions and `SambaNovaEmbeddings` for text embeddings. It requires setting `SAMBANOVA_API_KEY` (and optionally `SAMBANOVA_API_BASE` for SambaStack) as environment variables.

import os
from langchain_sambanova import ChatSambaNova
from langchain_core.messages import HumanMessage

# Ensure SAMBANOVA_API_KEY is set as an environment variable
# For SambaStack, SAMBANOVA_API_BASE might also be required.
os.environ['SAMBANOVA_API_KEY'] = os.environ.get('SAMBANOVA_API_KEY', 'your_sambanova_api_key_here')

# Initialize the SambaNova Chat model
llm = ChatSambaNova(
    model="Meta-Llama-3.3-70B-Instruct",  # Replace with your desired model
    max_tokens=1024,
    temperature=0.7,
    top_p=0.01,
)

# Invoke the model with a message
messages = [
    ("system", "You are a helpful assistant that translates English to French."),
    ("human", "I love programming."),
]

response = llm.invoke(messages)
print(response.content)

# Example for embeddings
from langchain_sambanova import SambaNovaEmbeddings

embeddings = SambaNovaEmbeddings(model="E5-Mistral-7B-Instruct") # Replace with your desired embedding model

query_embedding = embeddings.embed_query("What is the meaning of life?")
print(f"Embedding snippet: {str(query_embedding)[:50]}...")

view raw JSON →