Background LLM Polling

Interfacing with massive machine-learning models usually forces your servers to freeze while waiting for giant JSON payloads to generate and return. Even standard streaming setups can tax the main loop with parsing work.

Sofuu avoids this with native worker threads. Calling sofuu.ai.stream() hands the request to a background polling layer that parses Server-Sent Events as they arrive, leaving your entire JavaScript runtime free to handle other traffic.

// stream() runs on native worker threads — the JS loop never blocks
const stream = sofuu.ai.stream("Explain gravity.");
// Main logic loop continues seamlessly here!
// The iterator bridges native SSE chunks without freezing the Node loop
for await (const chunk of stream) {
  res.write(chunk.text);
}