Use Vercel AI SDK with SaladCloud

Build JavaScript and Next.js applications with AI SDK 7, the SaladCloud provider, and Salad AI Gateway or a self-hosted model.

Last Updated: September 17, 2026

Introduction

The Vercel AI SDK provides a shared interface for building AI-powered JavaScript and TypeScript applications. For AI SDK 7, use the published SaladCloud provider to connect to Salad AI Gateway. For self-hosted OpenAI-compatible servers, use @ai-sdk/openai-compatible instead.

This is a direct connection to SaladCloud using the AI SDK library, not a connection through the hosted Vercel AI Gateway. You do not need a Vercel account or Vercel API key.

The Vercel AI SDK works with SaladCloud in two ways:

• Salad AI Gateway - no infrastructure to deploy or manage. Available in beta with pay-per-token billing. Connect to the shared endpoint using your organization-specific AI Gateway API key.
• Self-hosted model - deploy your own SaladCloud container group for full control over the model, hardware, and configuration. Still very easy to set up and use.

Prerequisites

Before getting started, make sure you have:

• A SaladCloud account
• Node.js 22 or newer installed

Step-by-Step Setup

Step 1: Choose Your Backend

Salad AI Gateway

Salad AI Gateway is the fastest way to get started - no container groups to deploy, no cold starts to wait for.

1. Sign in to the SaladCloud portal.
2. Follow Pay-Per-Token Onboarding to add credits and create your organization’s AI Gateway API key.

This guide uses qwen3.6-35b-a3b.

Self-Hosted on SaladCloud

First, deploy an OpenAI-compatible LLM server on SaladCloud.

• Go to the SaladCloud portal and create an account if you do not already have one.
• Create an organization or choose an existing one, then click “Deploy a container group”.
• Select an LLM recipe. The Qwen3.6-35B-A3B (llama.cpp) recipe is a strong choice for general-purpose app development. On the recipe page, provide a name and deploy - the rest is preconfigured with recommended settings.
• Once deployed, your endpoint will be live and serving an OpenAI-compatible API.

Available recipes:

Ready-to-deploy recipes (best for less technical users):

• qwen3.6-35B-A3B - A powerful Mixture of Experts model optimized for instruction-following tasks.

Recipes for custom deployments (best for advanced users):

• llama.cpp - Supports GGUF models
• sglang - High-performance inference
• vllm - Popular LLM serving framework
• ollama - Simple model management
• tgi - Hugging Face Text Generation Inference server

After deployment, note your API endpoint URL (e.g., https://your-endpoint.salad.cloud).

Step 2: Install the Dependencies

In a new project directory, initialize an ES module project. If you already have a project, keep its existing configuration and adapt the examples to its module system.

npm init -y
npm pkg set type=module

Install the packages for your chosen backend:

Salad AI Gateway

npm install ai@7 @saladtechnologies-oss/ai-sdk-provider@^0.2.0

Version 0.2.0 of the SaladCloud provider implements the AI SDK 7 provider interface. Do not use it with ai@6 or older.

Self-Hosted on SaladCloud

npm install ai@7 @ai-sdk/openai-compatible@3

Step 3: Configure the SaladCloud Provider

Salad AI Gateway

Create a .env file with your organization’s AI Gateway API key:

SALAD_CLOUD_API_KEY=your-ai-gateway-api-key

Create provider.js:

provider.js

import { saladCloud } from '@saladtechnologies-oss/ai-sdk-provider'

export const model = saladCloud('qwen3.6-35b-a3b')

The provider reads SALAD_CLOUD_API_KEY and defaults to https://ai.salad.cloud/v1. You do not need to supply a base URL or custom headers. A regular SaladCloud API key is not a substitute for an AI Gateway key.

Self-Hosted on SaladCloud

Create a .env file for your container group:

SALAD_API_KEY=your-saladcloud-api-key
SALAD_BASE_URL=https://your-endpoint.salad.cloud/v1

Create provider.js. This example assumes Container Gateway authentication is enabled; omit headers if your deployment does not use it. If your model server requires additional authentication, configure that separately.

provider.js

import { createOpenAICompatible } from '@ai-sdk/openai-compatible'

const saladcloud = createOpenAICompatible({
  name: 'saladcloud',
  baseURL: process.env.SALAD_BASE_URL,
  headers: {
    'Salad-Api-Key': process.env.SALAD_API_KEY,
  },
})

export const model = saladcloud('qwen3.6-35b-a3b')

Replace the model ID with the one served by your deployment. Unlike AI Gateway, Container Gateway uses the Salad-Api-Key header when authentication is enabled.

Add .env and .env.local to your project’s .gitignore. Keep the key on the server: do not expose it in browser code or a NEXT_PUBLIC_ environment variable. In Next.js, run the provider inside a server-side route or action and store SALAD_CLOUD_API_KEY in .env.local or your hosting platform’s secret settings.

Step 4: Generate Text and Test the Connection

Create test.js. This example works with either provider.js above:

test.js

import { generateText } from 'ai'
import { model } from './provider.js'

const { text } = await generateText({
  model,
  prompt: 'Explain distributed GPU computing in one paragraph.',
})

console.log(text)

Run it with:

node --env-file=.env test.js

If you see a response, your setup is complete.

Streaming Chat Responses

Create stream.js to display text as it is generated:

stream.js

import { streamText } from 'ai'
import { model } from './provider.js'

const result = streamText({
  model,
  instructions: 'You are a helpful coding assistant.',
  prompt: 'How do I read a file in Python?',
})

for await (const chunk of result.textStream) {
  process.stdout.write(chunk)
}
process.stdout.write('\n')

node --env-file=.env stream.js

Troubleshooting

• Unsupported model version: check npm ls ai @saladtechnologies-oss/ai-sdk-provider. Use ai@7 with SaladCloud provider 0.2.0.
• Missing API key or HTTP 401: load .env with --env-file and use an organization-specific AI Gateway key in SALAD_CLOUD_API_KEY. For self-hosted deployments, check the separate Salad-Api-Key header configuration.
• Unexpected endpoint: remove an old SALAD_CLOUD_BASE_URL override if you intend to use the shared AI Gateway.