Vercel AI SDK - Neosantara Docs

The Vercel AI SDK is a unified TypeScript library for building reactive AI web applications. It supports token streaming, conversational UI hooks, and structured output extraction.

Setup

Install the core SDK and the OpenAI-compatible adapter:

npm i ai @ai-sdk/openai-compatible

pnpm add ai @ai-sdk/openai-compatible

bun add ai @ai-sdk/openai-compatible

export NEOSANTARA_API_KEY="nsk_your_api_key_here"

NEOSANTARA_API_KEY=nsk_your_api_key_here

$env:NEOSANTARA_API_KEY="nsk_your_api_key_here"

Provider Initialization

Instantiate the Neosantara provider using createOpenAICompatible:

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

export const neosantara = createOpenAICompatible({
  name: 'neosantara',
  apiKey: process.env.NEOSANTARA_API_KEY,
  baseURL: 'https://api.neosantara.xyz/v1',
});

Text Generation and Streaming

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

const neosantara = createOpenAICompatible({
  name: 'neosantara',
  apiKey: process.env.NEOSANTARA_API_KEY,
  baseURL: 'https://api.neosantara.xyz/v1',
});

const { text, usage } = await generateText({
  model: neosantara('gemini-3.8-flash'),
  prompt: 'Explain microservices architecture in 3 concise sentences.',
});

console.log(text);
console.log('Token usage:', usage);

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

const neosantara = createOpenAICompatible({
  name: 'neosantara',
  apiKey: process.env.NEOSANTARA_API_KEY,
  baseURL: 'https://api.neosantara.xyz/v1',
});

const result = streamText({
  model: neosantara('deepseek-v4.1-flash'),
  prompt: 'Provide 3 core clean architecture principles for TypeScript backends.',
});

for await (const delta of result.textStream) {
  process.stdout.write(delta);
}
console.log();

Next.js App Router Route (`app/api/chat/route.ts`)

Return a standard data stream response compatible with the frontend useChat hook:

app/api/chat/route.ts

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

const neosantara = createOpenAICompatible({
  name: 'neosantara',
  apiKey: process.env.NEOSANTARA_API_KEY,
  baseURL: 'https://api.neosantara.xyz/v1',
});

export async function POST(req: Request) {
  const { messages } = await req.json();

  const result = streamText({
    model: neosantara('deepseek-v4.1-flash'),
    messages,
  });

  return result.toDataStreamResponse();
}

Vercel AI SDK Method Comparison

Method	Return Type	Common Use Cases
`generateText`	Full text Promise	Background processing, batch jobs, webhook workers
`streamText`	Async text stream	Chat completion endpoints, Next.js streaming UIs
`generateObject`	Zod-validated Object	Strict JSON parsing, automated entity extraction

Next Steps

Related topics

Streaming Responses Chat Completions Model Catalog Integrations Overview

⌘I