Vercel AI SDK | OpenDunes Docs

The Vercel AI SDK ships a first-class OpenAI provider that accepts a custom baseURL. Swap it for OpenDunes and you gain access to the full model catalog while billing stays in Algerian Dinars.

Installation#

Shell

npm install ai @ai-sdk/openai

Provider Setup#

Create the provider once and reuse it across your app.

TypeScript

import { createOpenAI } from "@ai-sdk/openai";

export const opendunes = createOpenAI({
  baseURL: "https://opendunes.com/api/v1",
  apiKey: process.env.OPENDUNES_API_KEY,
});

Text Generation#

TypeScript

import { generateText } from "ai";
import { opendunes } from "@/lib/opendunes";

const { text } = await generateText({
  model: opendunes("anthropic/claude-sonnet-5"),
  prompt: "Explain micro-DA in one sentence.",
});

console.log(text);

Streaming in a Route Handler#

TypeScript

// app/api/chat/route.ts
import { streamText } from "ai";
import { opendunes } from "@/lib/opendunes";

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

  const result = streamText({
    model: opendunes("google/gemini-2.5-pro"),
    messages,
  });

  return result.toDataStreamResponse();
}

useChat Hook#

Pair the route handler above with the useChat hook on the client:

Tsx

"use client";

import { useChat } from "ai/react";

export function Chat() {
  const { messages, input, handleInputChange, handleSubmit } = useChat({
    api: "/api/chat",
  });

  return (
    <div>
      {messages.map((m) => (
        <p key={m.id}>
          <strong>{m.role}:</strong> {m.content}
        </p>
      ))}
      <form onSubmit={handleSubmit}>
        <input value={input} onChange={handleInputChange} />
        <button type="submit">Send</button>
      </form>
    </div>
  );
}

Switching Models#

Because opendunes("...") takes a slug string, you can switch models at runtime:

TypeScript

const model = opendunes(
  userTier === "pro" ? "anthropic/claude-opus-4.8" : "meta-llama/llama-4-maverick"
);

Browse the full catalog at /models or fetch it programmatically from GET /api/models.

Environment Variable#

Add to .env.local:

Shell

OPENDUNES_API_KEY=sk-your-key-here

Note. OPENDUNES_API_KEY is a server-side variable. Do not prefix it with NEXT_PUBLIC_ — it must never be exposed to the browser.

import { generateText } from "ai"; import { opendunes } from "@/lib/opendunes"; const { text } = await generateText({ model: opendunes("anthropic/claude-sonnet-5"), prompt: "Explain micro-DA in one sentence.", }); console.log(text);

Streaming in a Route Handler#

TypeScript

// app/api/chat/route.ts
import { streamText } from "ai";
import { opendunes } from "@/lib/opendunes";

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

  const result = streamText({
    model: opendunes("google/gemini-2.5-pro"),
    messages,
  });

  return result.toDataStreamResponse();
}

useChat Hook#

Pair the route handler above with the useChat hook on the client:

Tsx

"use client";

import { useChat } from "ai/react";

export function Chat() {
  const { messages, input, handleInputChange, handleSubmit } = useChat({
    api: "/api/chat",
  });

  return (
    <div>
      {messages.map((m) => (
        <p key={m.id}>
          <strong>{m.role}:</strong> {m.content}
        </p>
      ))}
      <form onSubmit={handleSubmit}>
        <input value={input} onChange={handleInputChange} />
        <button type="submit">Send</button>
      </form>
    </div>
  );
}