auto_mode

Asynchronously try each request, falling back when one with images is rejected.

Try each request in order, falling back when one with images is rejected.

Complete input for one OpenAIDecisions invocation.

Keeping the input and questions together in the Runnable input ensures both participate in composition, batching, and tracing.

Typed answers and metadata returned from one Decisions request.

Access every answer through answers, or use predicates, choices, scores, and refusals for views filtered by answer type. Each mapping is keyed by the question names supplied in DecisionRequest.questions.

Ask typed questions about text or images with the OpenAI Decisions API.

A single request can combine Predicate probabilities, Choice selections, and Score ratings over shared input. The response preserves probabilities, confidence, and token usage so application code can decide whether to act, route, or request human review.

Sync and async OpenAI clients are created on first use of invoke or ainvoke respectively.

Estimate the probability that a condition is true.

from langchain_openai.decisions import OpenAIDecisions, Predicate

decisions = OpenAIDecisions(model="gpt-6-luna")
response = decisions.invoke(
    {
        "input": "Production is down. Please help immediately.",
        "questions": {
            "urgent": Predicate(instructions="Is this request urgent?"),
        },
    }
)
if response.predicates["urgent"].probability >= 0.8:
    page_on_call_engineer()

Allow low-risk tool calls and block risky calls using OpenAI Decisions.

Intercepts explicitly configured tools immediately before execution and asks a Predicate question for the probability that each call is risky or insufficiently authorized. Calls below 0.5 execute normally. Other calls return an error ToolMessage without invoking the tool. Tools not listed in tools bypass classification.

The model receives the proposed tool call and up to 30 recent messages. Base64 images are included; audio, video, and file content, which the Decisions API does not accept, are replaced by placeholders. If a request with images is rejected (HTTP 400 or 413), the call is classified once more with images replaced by placeholders. Only explicit user messages authorize execution.

The middleware fails closed: refusals block the call, and classification errors propagate without executing the tool. It blocks risky calls; it does not request human approval.