Streaming

Set stream: true on either completion endpoint. The response uses text/event-stream; do not call response.json().

const response = await fetch("https://api.strixgate.dev/v1/chat/completions", options);
if (!response.ok || !response.body) throw new Error(`Request failed: ${response.status}`);

const reader = response.body.getReader();
const decoder = new TextDecoder();
for (;;) {
  const { done, value } = await reader.read();
  if (done) break;
  console.log(decoder.decode(value, { stream: true }));
}

Chunks may split across network reads. Production parsers should buffer text until a complete SSE event delimiter is available, then parse each data: line. OpenAI-compatible streams terminate with the protocol's completion sentinel. Anthropic requests are translated to Anthropic event names and payload shapes.

Headers arrive before the body and can be inspected immediately. Browser code can read the exposed x-strix-* routing headers on completion requests.