Getting Started with Salad AI Gateway

Last Updated: October 19, 2026

Salad AI Gateway is currently in beta. Pay-per-token access is self-service.

Prerequisites

A SaladCloud organization with a positive credit balance
An organization-specific AI Gateway API key. Follow Pay-Per-Token Onboarding to create one.

Base URL

All requests go to:

https://ai.salad.cloud/v1

Authentication

Pass your organization-specific AI Gateway API key as a Bearer token in the Authorization header. Usage is billed to the organization associated with that key:

Authorization: Bearer <your-ai-gateway-api-key>

Use the key created on your organization’s AI Gateway page, not the account’s Current API Key. The Salad-Api-Key header used by other SaladCloud APIs does not authenticate AI Gateway requests.

Store your key in the environment before running any example below:

export SALAD_AI_GATEWAY_API_KEY="<your-ai-gateway-api-key>"

Keep the key server-side and out of source control. See Manage Your AI Gateway API Key to copy, regenerate, or delete it.

Step 1: Test with curl

The quickest way to verify access is a single curl call:

curl https://ai.salad.cloud/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $SALAD_AI_GATEWAY_API_KEY" \
  -d '{
    "model": "qwen3.6-35b-a3b",
    "messages": [
      {"role": "user", "content": "Say hello in one sentence."}
    ]
  }'

You should receive a JSON response containing the model’s reply.

Step 2: Use the SaladCloud AI SDK Provider

For JavaScript and TypeScript applications built with the Vercel AI SDK, use the official @saladtechnologies-oss/ai-sdk-provider. The provider requires Node.js 22 or later and automatically uses the Salad AI Gateway base URL. Install the provider, Vercel AI SDK, and Zod:

npm install @saladtechnologies-oss/ai-sdk-provider ai zod

The provider reads your key from SALAD_CLOUD_API_KEY. If you followed the pay-per-token onboarding guide, map the existing variable before running your application:

export SALAD_CLOUD_API_KEY="$SALAD_AI_GATEWAY_API_KEY"

Create index.mjs:

index.mjs

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

const result = await generateText({
  model: saladCloud('qwen3.6-35b-a3b'),
  prompt: 'Say hello in one sentence.',
})

console.log(result.text)

Run the example:

node index.mjs

The provider also supports streaming, tool calling, structured output, reasoning options, and image inputs. See the SaladCloud AI SDK Provider repository for configuration and additional examples.

Step 3: Use the OpenAI Python SDK

Salad AI Gateway is fully compatible with the OpenAI Python SDK — just point it at a different base URL:

pip install openai

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://ai.salad.cloud/v1",
    api_key=os.environ["SALAD_AI_GATEWAY_API_KEY"],
)

response = client.chat.completions.create(
    model="qwen3.6-35b-a3b",
    messages=[{"role": "user", "content": "Explain what Salad AI Gateway is in two sentences."}],
)

print(response.choices[0].message.content)

Step 4: Streaming

For real-time output, enable streaming:

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://ai.salad.cloud/v1",
    api_key=os.environ["SALAD_AI_GATEWAY_API_KEY"],
)

stream = client.chat.completions.create(
    model="qwen3.6-35b-a3b",
    messages=[{"role": "user", "content": "Write a short poem about distributed computing."}],
    stream=True,
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)

Step 5: Use the OpenAI JavaScript/TypeScript SDK

npm install openai

import OpenAI from 'openai'

const client = new OpenAI({
  baseURL: 'https://ai.salad.cloud/v1',
  apiKey: process.env.SALAD_AI_GATEWAY_API_KEY,
})

const response = await client.chat.completions.create({
  model: 'qwen3.6-35b-a3b',
  messages: [{ role: 'user', content: 'Hello from Salad AI Gateway!' }],
})

console.log(response.choices[0].message.content)

Available Models

Model	Context Window	Best For
`qwen3.6-35b-a3b`	262,144 tokens	Complex reasoning, agentic tasks, coding
`qwen3.6-27b`	195,200 tokens	General purpose, balanced speed and capability
`qwen3.5-35b-a3b`	262,144 tokens	Cost-efficient agentic tasks
`qwen3.5-9b`	196,608 tokens	High-volume, simple queries

See the Models Reference for full details.

Next Steps

Browse integration guides for agentic tools, IDEs, and frameworks
Read the Models Reference for model capabilities and limits
Review AI Gateway Pricing for input and output token rates
Check AI Gateway Errors for credit or access failures

AI Gateway

How-to Guides

Reference

Integrations

Getting Started with Salad AI Gateway

Prerequisites

Base URL

Authentication

Step 1: Test with curl

Step 2: Use the SaladCloud AI SDK Provider

Step 3: Use the OpenAI Python SDK

Step 4: Streaming

Step 5: Use the OpenAI JavaScript/TypeScript SDK

Available Models

Next Steps

​Prerequisites

​Base URL

​Authentication

​Step 1: Test with curl

​Step 2: Use the SaladCloud AI SDK Provider

​Step 3: Use the OpenAI Python SDK

​Step 4: Streaming

​Step 5: Use the OpenAI JavaScript/TypeScript SDK

​Available Models

​Next Steps

Prerequisites

Base URL

Authentication

Step 1: Test with curl

Step 2: Use the SaladCloud AI SDK Provider

Step 3: Use the OpenAI Python SDK

Step 4: Streaming

Step 5: Use the OpenAI JavaScript/TypeScript SDK

Available Models

Next Steps