🦋 New version release

[brentrager]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

@smooai/smooth-operator-core@1.11.0

Minor Changes

• 875b7a6: Ask the gateway for usage on every streaming request (stream_options: {"include_usage": true}), in all
  five engines.

  The OpenAI streaming API omits usage unless it is explicitly requested, and nothing here ever
  requested it. So the missing usage chunk was never the gateway losing data — it was the gateway
  correctly honouring a request that never asked. Everything built on top of that misattribution
  compensates for one unset request parameter: two char-count estimators, prompt_tokens hardcoded to
  0, completion_tokens = content.len() / 4, the usage_estimated and cost_estimated flags, and a
  cross-language "is this measured?" convention.

  Verified against llm.smoo.ai (LiteLLM 1.95.0, groq-gpt-oss-120b), same prompt both ways:

  • without the field — 7 chunks, 0 carrying usage
  • with the field — 8 chunks, 1 carrying "prompt_tokens": 73, "completion_tokens": 8

  Sent only when streaming: it is meaningless otherwise, and leaving it off keeps a non-streaming
  request byte-identical to before. Python honours an explicit caller-supplied stream_options rather
  than overriding it.

  Not fixed by this, and worth knowing: the per-request cost headers are present on a streamed
  response but all read 0.0, because at header-flush time the completion has not been priced yet.
  parse_gateway_cost already maps 0 to None, so gateway cost stays unavailable on the streaming
  path and the response_id → LiteLLM_SpendLogs.request_id join remains the only authoritative
  per-turn cost there.

@smooai/smooth-operator-temporal@1.9.4

Patch Changes

• Updated dependencies [875b7a6]
  • @smooai/smooth-operator-core@1.11.0