Streaming

Pass stream: true to receive Server-Sent Events. The same SDK iterators that work with OpenAI work here.

stream = client.chat.completions.create(
  model="sable-llama-3.3-70b",
  messages=[{"role":"user","content":"Stream me a haiku about privacy."}],
  stream=True,
)
for chunk in stream:
  delta = chunk.choices[0].delta.content or ""
  print(delta, end="", flush=True)

Failover behavior

If the primary upstream fails before the first byte streams, Sable transparently fails over to the next configured upstream. Once bytes start flowing, there's no failover — interrupted streams surface as an SSE disconnect.