sambanova-ai-provider

Vercel AI Provider for running LLMs locally using SambaNova's models.

Table of Contents

• Requirements
• Installation
• Setup Environment
• Provider Instance
• Models
  • Tested models and capabilities
  • Image input
  • Tool calling
  • Embeddings
• Examples
• Intercepting Fetch requests

Requirements

API key can be obtained from the SambaNova Cloud Platform.

Installation

The SambaNova provider is available in the sambanova-ai-provider module. You can install it with

npm:

npm install sambanova-ai-provider

yarn:

yarn add sambanova-ai-provider

or pnpm:

pnpm add sambanova-ai-provider

Setup Environment

You will need to setup a SAMBANOVA_API_KEY environment variable. You can get your API key on the SambaNova Cloud Portal.

Provider Instance

You can import the default provider instance sambanova from sambanova-ai-provider:

import { sambanova } from 'sambanova-ai-provider';

If you need a customized setup, you can import createSambaNova from sambanova-ai-provider and create a provider instance with your settings:

import { createSambaNova } from 'sambanova-ai-provider';

const sambanova = createSambaNova({
  apiKey: 'YOUR_API_KEY',
  // Optional settings
});

You can use the following optional settings to customize the SambaNova provider instance:

• baseURL string

  Use a different URL prefix for API calls, e.g. to use proxy servers. The default prefix is https://api.sambanova.ai/v1.

• apiKey string

  API key that is being sent using the Authorization header. It defaults to the SAMBANOVA_API_KEY environment variable*.

• headers Record<string,string>

  Custom headers to include in the requests.

• fetch (input: RequestInfo, init?: RequestInit) => Promise<Response>

  Custom fetch implementation. Defaults to the global fetch function. You can use it as a middleware to intercept requests, or to provide a custom fetch implementation for e.g. testing.

* If you set the environment variable in a .env file, you will need to use a loader like dotenv in order for the script to read it.

Models

You can use SambaNova models on the provider instance. The first argument is the model ID, e.g. Meta-Llama-3.3-70B-Instruct.

const model = sambanova('Meta-Llama-3.3-70B-Instruct');

Tested models and capabilities

This provider is capable of generating and streaming text, interpreting image inputs, run tool callings, and use embeddings.

At least it has been tested with the following features:

Chat completion	Image input	Tool calling	Embeddings
✅	✅	✅	✅

Image input

You need to use any of the following models for visual understanding:

• Llama-4-Maverick-17B-128E-Instruct
• Llama-4-Scout-17B-16E-Instruct

SambaNova vision models support up to five (5) images per request. They don't support URLs.

Tool calling

You can use any of the Function calling supported models for tool calling.

Embeddings

You can use the E5-Mistral-7B-Instruct model to use the embeddings feature of the SambaNova provider.

Examples

On the examples folder you will find some Markdown files containing simple code snippets of some of the features of the SambaNova Provider.

Intercepting Fetch requests

This provider supports Intercepting Fetch Requests.