Stream a model response

Set stream: true on a chat-completions request and consume incremental choices. Keep cancellation under your application's request lifecycle and handle a partial response as a partial response.

const stream = await openai.chat.completions.create({
  model: "openpond-chat",
  stream: true,
  messages: [{ role: "user", content: "Explain streaming in one paragraph." }],
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

The selected model and endpoint determine supported request options. Verify the current model response rather than assuming all OpenAI features are available.

Next: Tool calling.