Requests go to POST /openai/v1/decisions, use a workspace-scoped LangSmith API key, and require an OPENAI_API_KEY provider secret. The model is named using its native OpenAI name, without the openai/ prefix. The gateway preserves the API’s native request and response formats. Its existing access, rate-limit, and budget policies apply; token use counts toward spend caps, and requests are traced to LangSmith with inputs, answers, and usage recorded. The documentation illustrates a predicate question and a response containing a probability value. It does not establish broader performance or reliability claims. Source: https://docs.langchain.com/langsmith/llm-gateway-decision-models
Raw SignalOct 9, 20261 min read1 source
LangSmith LLM Gateway now supports the OpenAI Decisions API.
In one minute
LangSmith’s documentation now describes support for the OpenAI Decisions API through the LLM Gateway.