# Decisions API

<!-- WARNING: THIS FILE WAS AUTOGENERATED! DO NOT EDIT! -->

The Decisions API evaluates text or image input against one or more questions and returns structured results. Use it to route requests, classify content, assign priority, or choose the next step in an application workflow.

## Key features

-   Returns a probability for a true-or-false condition
-   Selects from caller-defined choices and returns their probabilities
-   Scores input against ordered levels
-   Evaluates multiple independent questions against the same input
-   Accepts text, images, or combined text and image input
-   Returns completed results in one synchronous HTTP response

## Decision types

| Type        | Use it to                                          | Result                                                                    |
| ----------- | -------------------------------------------------- | ------------------------------------------------------------------------- |
| `predicate` | Check whether a condition about the input is true. | A `probability` from `0` to `1`.                                          |
| `choice`    | Select one of the supplied options.                | A `choice` and a probability distribution over the options.               |
| `score`     | Rate the input against ordered levels.             | A probability-weighted `score` and a distribution over the level indices. |

For example, an application can evaluate a customer request to determine whether it needs escalation, which team should handle it, and how urgent it is.

## Models

| Model                | Description                                                                                           |
| -------------------- | ----------------------------------------------------------------------------------------------------- |
| `zoom-decisions`     | Default model, optimized for latency and cost.                                                        |
| `zoom-decisions-pro` | Optimized for higher decision accuracy and more complex workloads. Select it explicitly with `model`. |

Model access and pricing depend on your AI Services entitlement. The model identifiers represent service tiers, rather than the underlying model implementation.

## Make a decision

1. Provide the shared evidence in `input`.
2. Add one or more typed `questions`, each with a unique `name` and instructions describing what to evaluate.
3. Send a Fast mode request.
4. Read the corresponding entries in `answers` and apply your application's decision rules.

Questions in one request are independent. When a question depends on an earlier answer, submit it in a separate request after receiving that answer.

For the endpoint, request fields, and response handling, see [Fast mode](/docs/ai-services/decisions/fast-mode/).

## Image input

Images must be inline data URLs in `input_image` content parts. Set `enable_image_input` to `true` for any request containing images. Image input is disabled by default.

Image-specific safety guardrails are not available. General AI Services security, privacy, and request validation controls apply.

## Processing mode

Decisions supports synchronous Fast mode only. It does not provide Batch jobs, streaming, or WebSocket execution. The API evaluates the supplied evidence and returns results; it does not execute tools or take actions on your application's behalf.

## Resources

AI Services are part of the [Zoom Build platform](/docs/build/). Use [Zoom.ai](https://zoom.ai/) to create a project, get credentials, and manage your AI Services plan.