Vercel AI SDK
An open-source TypeScript toolkit for building AI apps with streaming and a unified model API.
Makes streaming, tool calls and swapping models a one-liner. If you're building AI features in React or Next, this is the layer to start from — free and open source.
The AI SDK is an open-source TypeScript toolkit for building AI features into web apps. It gives you one unified API across model providers, first-class streaming, tool calling and structured output, and tight integration with React, Next, Svelte and Vue. It is the layer that turns "call an LLM" into a few readable lines.
Start with this
import { streamText } from 'ai';
import { openai } from '@ai-sdk/openai';
const result = streamText({
model: openai('gpt-4o'),
prompt: 'Explain React Server Components in one paragraph.',
});
// stream tokens to the client as they arrive
for await (const chunk of result.textStream) {
process.stdout.write(chunk);
}The take, for AI builders
For AI builders this is the layer to start from: streaming, tool calls and structured output are one-liners, and swapping providers is a single import change, so you're never locked to one model. It pairs naturally with React/Next, and an agent can scaffold a working streaming chat with it in minutes. Reach past it to a heavier framework only when you genuinely need complex multi-step agent orchestration; for most app features, this is all you need.
When to reach for it
✓ Good fit
- Building AI features in a React, Next, Svelte or Vue app
- You want streaming, tool calls and structured output without boilerplate
- You want to swap model providers without rewriting code
✕ Not for
- You need heavy multi-agent orchestration and graphs (look at LangChain)
- You're not working in JavaScript or TypeScript
Alternatives
A heavier, batteries-included framework when your agent pipeline is genuinely complex.
If you just need multi-provider model routing at the API level, not an app SDK.
Related in AI APIs & models
Anthropic API
The API for the Claude family of models, with long context, tool use and prompt caching.
- llm
- api
- claude
Strong at coding, careful reasoning and long context — and the engine behind a lot of the agent tools in this library. Pay-per-token; prompt caching keeps repeat costs down.
Visit ↗ElevenLabs
A voice AI platform for realistic text-to-speech, voice cloning and dubbing via API.
- voice
- text-to-speech
- api
The best-sounding text-to-speech and voice cloning around, with a clean API. The default when your app needs a voice that doesn't sound like a robot.
Visit ↗fal
An inference platform optimised for fast generative media models like image and video.
- inference
- image
- video
The go-to for fast image and video generation behind an API. If you're building a media-gen feature, its speed and model selection are hard to beat; you pay per use.
Visit ↗