Skip to main content
Use streamText when you want partial tokens as they arrive. 
import { aiStats } from "@ai-stats/ai-sdk-provider";
import { streamText } from "ai";

const result = streamText({
  model: aiStats("openai/gpt-5-nano"),
  prompt: "Explain SSE streaming in plain language.",
});

for await (const chunk of result.textStream) {
  process.stdout.write(chunk);
}

​
Streaming with messages

import { aiStats } from "@ai-stats/ai-sdk-provider";
import { streamText } from "ai";

const result = streamText({
  model: aiStats("openai/gpt-5-nano"),
  messages: [
    { role: "user", content: "Write a five-step rollout plan for a gateway migration." },
  ],
});

let text = "";
for await (const chunk of result.textStream) {
  text += chunk;
}

console.log(text);

​
Production tips

  • Prefer streaming for long responses and interactive UIs.
  • Handle cancellation/timeout in your runtime for abandoned streams.
  • Keep model IDs fresh via /v1/gateway/models to avoid stale aliases.
Last modified on March 16, 2026