> ## Documentation Index
> Fetch the complete documentation index at: /docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Create a monitor

> Create a monitor on a project: `TRACING` on an LLM tracing project, or
`DATA_QUALITY` on an ML project. The monitor is published and starts
evaluating immediately, and the response matches
`GET /v2/monitors/{monitor_id}`.

**Payload Requirements**
- `type`, `project_id`, `name`, `metric`, `dimension` and `threshold`
  are required. Field-level limits are documented on each field.
- `project_id` must be a project of the kind `type` needs that is not
  an Arize demo project.
- `dimension.name` and each filter's name must match the project's
  data exactly: recent tracing data for `TRACING`, the monitored
  `environment` for `DATA_QUALITY`. A `CUSTOM_BASELINE` reads
  production data, so its filters and the monitored dimension must
  also be in production data. Embedding and dictionary fields cannot
  be monitored or filtered.
- `metric` must suit the dimension; each request's `metric` field
  lists the metrics each kind of dimension supports.
- `DATA_QUALITY` monitors require `baseline_config` for `NEW_VALUES`
  and `MISSING_VALUES` and reject it otherwise. `MODEL_BASELINE`
  requires a primary baseline, and every `model_versions` entry must
  be a version of the project.
- `INTEGRATION` and `WEBHOOK` destinations require a plan with alert
  integrations and must belong to the project's organization.
- Unknown fields, `null` values and system-managed fields (`id`,
  `status`, `created_at` and so on) are rejected.

Structural problems return `400`, and values that break a rule above
return `422`, including a project of the wrong kind for `type`. A
project that does not exist or that the caller cannot read returns
`404`. A caller who can read the project but cannot create monitors in
it gets `403` before the body is validated. A `5xx` does not guarantee
the monitor was not created, so list the project's monitors by `name`
before retrying.

**Valid example**
```json
{
  "type": "TRACING",
  "project_id": "TW9kZWw6MTIz",
  "name": "Root span latency p95",
  "metric": "P95",
  "dimension": { "category": "SPAN_ATTRIBUTE", "name": "latency_ms" },
  "threshold": { "type": "MANUAL_SINGLE", "operator": "GREATER_THAN", "value": 2000 },
  "evaluation_window_length_seconds": 3600,
  "filters": [
    {
      "dimension": { "category": "SPAN_ATTRIBUTE", "name": "parent_id" },
      "operator": "EQUALS",
      "values": ["NULL"]
    }
  ],
  "notification_configs": [
    { "type": "EMAIL", "email_address": "oncall@example.com" }
  ]
}
```

**Invalid example** (a system-managed field, and a range whose lower
bound uses an upper-bound operator)
```json
{
  "id": "TW9uaXRvcjo0NTY=",
  "type": "TRACING",
  "project_id": "TW9kZWw6MTIz",
  "name": "Hallucination rate",
  "metric": "AVG",
  "dimension": { "category": "LLM_EVAL", "name": "eval.hallucination.score" },
  "threshold": {
    "type": "MANUAL_RANGE",
    "lower": { "operator": "LESS_THAN", "value": 0.2 },
    "upper": { "operator": "LESS_THAN", "value": 0.8 }
  }
}
```

<Warning>This endpoint is in alpha, read more [here](https://arize.com/docs/ax/rest-reference#api-version-stages).</Warning>




## OpenAPI

````yaml https://api.arize.com/v2/spec.yaml post /v2/monitors
openapi: 3.0.3
info:
  title: Arize REST API
  version: 2.0.0
  description: |
    API specification for the backend data server. The API is hosted globally
    at https://api.arize.com/v2 or in your own environment.
  license:
    name: Apache-2.0
    url: https://www.apache.org/licenses/LICENSE-2.0
servers:
  - description: Global
    url: https://api.arize.com
  - description: Regional
    url: https://api.{region}.arize.com
    variables:
      region:
        default: eu-west-1a
        enum:
          - eu-west-1a
          - ca-central-1a
  - description: Custom Host
    url: https://{host}
    variables:
      host:
        default: api.arize.com
security:
  - bearerAuth: []
tags:
  - name: AI Integrations
    description: |
      AI integrations configure access to external LLM providers (e.g. OpenAI,
      Azure OpenAI, AWS Bedrock, Vertex AI). Integrations can be scoped to the
      entire account, a specific organization, or a specific space.
  - name: Alignment
    description: >
      Alignment measures agreement between evaluator labels and human
      annotations.

      Tasks define the dataset columns and comparison function. Runs record the

      resulting score for a dataset version.
  - name: Annotation Configs
    description: >
      Annotation configs allow you to define consistent annotation schemas that

      can be reused across your workspace, ensuring evaluations are structured
      and

      comparable over time.
  - name: Annotation Queues
    description: >
      Annotation queues help you organize and manage human evaluation workflows.

      Use queues to assign spans or examples to annotators for review and
      labeling.
  - name: API Keys
    description: >
      API keys are used to authenticate requests to the Arize API. List your
      keys

      to view metadata; the raw secret is never returned after creation.
  - name: Audit Logs
    description: >
      Audit logs record authenticated user actions within an account, providing
      a

      chronological trail for security and compliance review. Access requires

      account admin privileges and audit logging to be enabled.
  - name: Dashboards
    description: >
      Dashboards organize charts and summaries that help teams understand their
      data.
  - name: Datasets
    description: |
      Datasets are structured, version-controlled example collections you use to
      run, evaluate, and track LLM experiments.
  - name: Evaluators
    description: >
      Evaluators are reusable evaluation configurations used to assess the
      quality

      of LLM outputs. They can be template-based (using LLM judges) or
      code-based.
  - name: Experiments
    description: >
      Experiments let you systematically test prompt/model changes using
      datasets,

      tasks, and evaluators.
  - name: Integrations
    description: >
      Integrations configure access to external LLM providers (e.g. OpenAI,

      Azure OpenAI, AWS Bedrock, Vertex AI), notifications services (e.g.
      PagerDuty, Slack), and

      your own agents. Integrations can be scoped to the entire account, a
      specific

      organization, or a specific space.
  - name: Monitors
    description: >
      Monitors continuously track a metric over your model or LLM application
      data

      and alert you when it crosses a threshold. Each monitor watches a single

      metric - data quality, model performance, drift, a custom metric, or a

      tracing metric - and moves between statuses as the metric passes in and
      out of its 

      healthy range.
  - name: Organizations
    description: >
      Organizations are top-level containers within an Arize AX account for
      grouping spaces.
  - name: Projects
    description: |
      Projects represent LLM applications being monitored in Arize where you can
      observe traces and spans.
  - name: Prompts
    description: >
      Prompts are reusable, versioned templates for LLM interactions. Use
      prompts

      to standardize and manage how you interact with LLMs across your
      application.
  - name: Resource Restrictions
    description: |
      Endpoints for restricting and unrestricting resources (projects, models).
  - name: Role Bindings
    description: |
      Role bindings assign a role to a user on a resource. REST currently
      supports space- and project-scoped bindings.
  - name: Roles
    description: >
      Roles define sets of permissions that can be assigned to users within an

      account. Create custom roles to tailor access control to your team's
      needs.
  - name: SAML
    description: Enterprise account-admin SAML configuration and role mappings.
  - name: Spaces
    description: >
      Spaces are containers within an organization for grouping related
      projects,

      datasets, and experiments, enabling collaboration or isolated
      experimentation

      with role-based access control.
  - name: Spans
    description: |
      Spans represent individual operations within a trace. A span captures the
      timing, status, and attributes of a single operation in your application.
  - name: Tags
    description: |
      Tags are reusable labels defined once per space and attached to resources
      across the platform, so the same vocabulary can be applied to projects,
      datasets, prompts, and more. Tags are shared: renaming or deleting a tag
      affects every resource it is attached to.
  - name: Tasks
    description: |
      Tasks are configurable units of work that tie one or more evaluators to a
      data source (project or dataset). Use tasks to automate evaluation of LLM
      outputs, with support for continuous evaluation and backfill runs.
  - name: Traces
    description: |
      A trace is the collection of spans sharing a trace ID, representing a
      single end-to-end request through an LLM application. Use the Traces
      endpoint to retrieve traces with all of their spans in one call.
  - name: Users
    description: >
      Users represent members of an account. The Users endpoints allow creating,

      listing, updating (display name), and removing users from the account
      programmatically.
  - name: Webhooks
    description: >
      Webhooks are organization-owned destinations that receive event deliveries

      over HTTPS. Deliveries are authenticated with a bearer token or signed
      with

      an HMAC signing secret — the secret is returned exactly once, when the

      webhook is created. Delivery attempts are recorded and can be listed for

      debugging. To choose which events a webhook receives, manage its

      subscriptions through the prompt and evaluator webhook-subscription

      endpoints.
paths:
  /v2/monitors:
    post:
      tags:
        - Monitors
      summary: Create a monitor
      description: >
        Create a monitor on a project: `TRACING` on an LLM tracing project, or

        `DATA_QUALITY` on an ML project. The monitor is published and starts

        evaluating immediately, and the response matches

        `GET /v2/monitors/{monitor_id}`.


        **Payload Requirements**

        - `type`, `project_id`, `name`, `metric`, `dimension` and `threshold`
          are required. Field-level limits are documented on each field.
        - `project_id` must be a project of the kind `type` needs that is not
          an Arize demo project.
        - `dimension.name` and each filter's name must match the project's
          data exactly: recent tracing data for `TRACING`, the monitored
          `environment` for `DATA_QUALITY`. A `CUSTOM_BASELINE` reads
          production data, so its filters and the monitored dimension must
          also be in production data. Embedding and dictionary fields cannot
          be monitored or filtered.
        - `metric` must suit the dimension; each request's `metric` field
          lists the metrics each kind of dimension supports.
        - `DATA_QUALITY` monitors require `baseline_config` for `NEW_VALUES`
          and `MISSING_VALUES` and reject it otherwise. `MODEL_BASELINE`
          requires a primary baseline, and every `model_versions` entry must
          be a version of the project.
        - `INTEGRATION` and `WEBHOOK` destinations require a plan with alert
          integrations and must belong to the project's organization.
        - Unknown fields, `null` values and system-managed fields (`id`,
          `status`, `created_at` and so on) are rejected.

        Structural problems return `400`, and values that break a rule above

        return `422`, including a project of the wrong kind for `type`. A

        project that does not exist or that the caller cannot read returns

        `404`. A caller who can read the project but cannot create monitors in

        it gets `403` before the body is validated. A `5xx` does not guarantee

        the monitor was not created, so list the project's monitors by `name`

        before retrying.


        **Valid example**

        ```json

        {
          "type": "TRACING",
          "project_id": "TW9kZWw6MTIz",
          "name": "Root span latency p95",
          "metric": "P95",
          "dimension": { "category": "SPAN_ATTRIBUTE", "name": "latency_ms" },
          "threshold": { "type": "MANUAL_SINGLE", "operator": "GREATER_THAN", "value": 2000 },
          "evaluation_window_length_seconds": 3600,
          "filters": [
            {
              "dimension": { "category": "SPAN_ATTRIBUTE", "name": "parent_id" },
              "operator": "EQUALS",
              "values": ["NULL"]
            }
          ],
          "notification_configs": [
            { "type": "EMAIL", "email_address": "oncall@example.com" }
          ]
        }

        ```


        **Invalid example** (a system-managed field, and a range whose lower

        bound uses an upper-bound operator)

        ```json

        {
          "id": "TW9uaXRvcjo0NTY=",
          "type": "TRACING",
          "project_id": "TW9kZWw6MTIz",
          "name": "Hallucination rate",
          "metric": "AVG",
          "dimension": { "category": "LLM_EVAL", "name": "eval.hallucination.score" },
          "threshold": {
            "type": "MANUAL_RANGE",
            "lower": { "operator": "LESS_THAN", "value": 0.2 },
            "upper": { "operator": "LESS_THAN", "value": 0.8 }
          }
        }

        ```


        <Warning>This endpoint is in alpha, read more
        [here](https://arize.com/docs/ax/rest-reference#api-version-stages).</Warning>
      operationId: create_monitor
      requestBody:
        $ref: '#/components/requestBodies/CreateMonitorRequestBody'
      responses:
        '201':
          $ref: '#/components/responses/Monitor'
        '400':
          $ref: '#/components/responses/BadRequest'
        '401':
          $ref: '#/components/responses/Unauthorized'
        '403':
          $ref: '#/components/responses/ForbiddenOrPlanUpgradeRequired'
        '404':
          $ref: '#/components/responses/NotFound'
        '422':
          $ref: '#/components/responses/UnprocessableEntity'
        '429':
          $ref: '#/components/responses/RateLimitExceeded'
components:
  requestBodies:
    CreateMonitorRequestBody:
      description: The monitor to create. `type` selects the variant.
      required: true
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/CreateMonitorRequest'
          examples:
            manual_single:
              summary: P95 latency of root spans above 2 seconds
              value:
                type: TRACING
                project_id: TW9kZWw6MTIz
                name: Root span latency p95
                metric: P95
                dimension:
                  category: SPAN_ATTRIBUTE
                  name: latency_ms
                threshold:
                  type: MANUAL_SINGLE
                  operator: GREATER_THAN
                  value: 2000
                evaluation_window_length_seconds: 3600
                filters:
                  - dimension:
                      category: SPAN_ATTRIBUTE
                      name: parent_id
                    operator: EQUALS
                    values:
                      - 'NULL'
                notification_configs:
                  - type: EMAIL
                    email_address: oncall@example.com
            data_quality_new_values:
              summary: >-
                New values of a feature compared with the previous week of
                production data
              value:
                type: DATA_QUALITY
                project_id: TW9kZWw6NDU2
                name: New merchant categories
                metric: NEW_VALUES
                dimension:
                  category: FEATURE_LABEL
                  name: merchant_category
                baseline_config:
                  type: CUSTOM_BASELINE
                  window:
                    type: MOVING
                    moving_window_seconds: 604800
                    moving_window_delay_seconds: 86400
                threshold:
                  type: MANUAL_SINGLE
                  operator: GREATER_THAN
                  value: 0
            dynamic_range_scheduled:
              summary: Eval score outside a deviation band, checked hourly on weekdays
              value:
                type: TRACING
                project_id: TW9kZWw6MTIz
                name: Correctness drift
                metric: AVG
                dimension:
                  category: LLM_EVAL
                  name: eval.correctness.score
                threshold:
                  type: DYNAMIC_RANGE
                  calculation: MAD
                  lower:
                    operator: GREATER_THAN_OR_EQUAL
                    multiplier: 2
                  upper:
                    operator: LESS_THAN_OR_EQUAL
                    multiplier: 2
                scheduled_runtime:
                  enabled: true
                  cadence_seconds: 3600
                  days_of_week:
                    - 1
                    - 2
                    - 3
                    - 4
                    - 5
                notification_configs:
                  - type: WEBHOOK
                    id: V2ViaG9vazo3
  responses:
    Monitor:
      description: Returns a single monitor object
      content:
        application/json:
          schema:
            $ref: '#/components/schemas/Monitor'
          example:
            id: TW9uaXRvcjo4NzEwMToxMlo5Wg==
            name: Test Monitor
            project_id: TW9kZWw6ODU0MjAxODQ6RFNxUg==
            uri: >-
              https://app.arize.com/organizations/ZXhhbXBsZU9yZ2FuaXphdGlvbjo1Njg6V2RnREk=/spaces/U3BhY2U6NzU1OnBsR1c=/monitors/TW9uaXRvcjo4NzEwMToxMlo5Wg==
            status: TRIGGERED
            created_at: '2026-07-09T22:07:34.010Z'
            updated_at: '2026-07-09T23:22:14.638Z'
            threshold:
              type: DYNAMIC_RANGE
              calculation: STDEV
              lower:
                operator: GREATER_THAN
                multiplier: 0.00001
              upper:
                operator: LESS_THAN_OR_EQUAL
                multiplier: 0.00001
            notifications_enabled: true
            manual_evaluation_enabled: false
            created_by_user_id: VXNlcjoxMTIzOlRlc3Rlcg==
            filters:
              - dimension:
                  category: LLM_EVAL
                  name: eval.Test Eval.label
                operator: CONTAINS_STRING
                values:
                  - No Value
                  - 'NULL'
              - dimension:
                  category: USER_ANNOTATION
                  name: annotation.Accuracy.score
                operator: EQUALS
                values:
                  - 'NULL'
            notification_configs:
              - type: EMAIL
                email_address: test.user1@example.com
              - type: EMAIL
                email_address: test.user2@example.com
              - type: EMAIL
                email_address: test.admin@example.com
              - type: WEBHOOK
                id: V2ViaG9vazoyMzpBczpH
                url: https://webhook.site/test-webhook-1
              - type: WEBHOOK
                id: V2ViaG9vazoyNDpTdW5x
                url: https://testsite.com/test5
            downtime:
              start: '2026-07-09T22:47:02.155Z'
              duration_seconds: 3600
              frequency_days: 1
            scheduled_runtime:
              enabled: true
              cadence_seconds: 10800
              days_of_week:
                - 0
            evaluation_window_length_seconds: 86400
            delay_seconds: 0
            evaluated_at: '2026-07-09T23:00:00.000Z'
            latest_computed_value: 251
            notes: >-
              This monitor should trigger when the metric moves slightly above
              the baseline.
            type: TRACING
            metric: COUNT
            dimension:
              category: SPAN_ATTRIBUTE
              name: attributes.output.value
    BadRequest:
      description: Invalid request
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/Problem'
          example:
            status: 400
            title: Invalid request parameters
            detail: The 'name' field is required and must be a non-empty string.
            instance: /resource
            type: https://arize.com/docs/ax/rest-reference/errors#invalid-request
    Unauthorized:
      description: Authentication is required
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/Problem'
          example:
            status: 401
            title: Authentication required
            detail: You must be authenticated to access this resource.
            instance: /resource
            type: >-
              https://arize.com/docs/ax/rest-reference/errors#authentication-required
    ForbiddenOrPlanUpgradeRequired:
      description: >-
        Insufficient permissions, the account's plan doesn't include this
        capability, or the account has reached its plan's limit
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/ForbiddenProblem'
          examples:
            forbidden:
              summary: Insufficient permissions
              value:
                status: 403
                title: Access forbidden
                detail: You do not have permission to access this resource.
                instance: /resource/12345
                type: >-
                  https://arize.com/docs/ax/rest-reference/errors#access-forbidden
            planUpgradeRequired:
              summary: Plan doesn't include the capability
              value:
                status: 403
                title: Plan Upgrade Required
                detail: Upgrade your plan to customize dashboards.
                type: >-
                  https://arize.com/docs/ax/rest-reference/errors#plan-upgrade-required
                entitlement: DASHBOARD_CUSTOMIZATION
                required_plan: PRO
            planLimitReached:
              summary: Account reached its plan's limit on a count
              value:
                status: 403
                title: Plan Upgrade Required
                detail: >-
                  Your plan's limit on spaces has been reached. Upgrade your
                  plan to create more spaces.
                type: >-
                  https://arize.com/docs/ax/rest-reference/errors#plan-upgrade-required
                entitlement: SPACES
                limit: 1
                current_count: 1
                required_plan: PRO
    NotFound:
      description: Not found
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/Problem'
          example:
            status: 404
            title: Resource not found
            detail: The requested resource with ID '12345' was not found.
            instance: /resource/12345
            type: https://arize.com/docs/ax/rest-reference/errors#resource-not-found
    UnprocessableEntity:
      description: Unprocessable entity
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/Problem'
          example:
            status: 422
            title: Unprocessable Entity
            detail: One or more fields failed validation.
            instance: /resource/12345
            type: >-
              https://arize.com/docs/ax/rest-reference/errors#unprocessable-entity
    RateLimitExceeded:
      description: Rate limit exceeded
      headers:
        Retry-After:
          description: |
            When throttled (429), how long to wait before retrying. Value is
            either a delta-seconds integer.
          schema:
            type: integer
            minimum: 0
          example: 42
      content:
        application/problem+json:
          schema:
            $ref: '#/components/schemas/Problem'
          example:
            status: 429
            title: Rate limit exceeded
            detail: >-
              You have exceeded the allowed number of requests. Please try again
              later.
            instance: /resource
            type: >-
              https://arize.com/docs/ax/rest-reference/errors#rate-limit-exceeded
  schemas:
    CreateMonitorRequest:
      oneOf:
        - $ref: '#/components/schemas/CreateTracingMonitorRequest'
        - $ref: '#/components/schemas/CreateDataQualityMonitorRequest'
      discriminator:
        propertyName: type
        mapping:
          TRACING: '#/components/schemas/CreateTracingMonitorRequest'
          DATA_QUALITY: '#/components/schemas/CreateDataQualityMonitorRequest'
      description: |
        The monitor to create. `type` selects the variant: `TRACING` on an LLM
        tracing project, or `DATA_QUALITY` on an ML project.
    Monitor:
      oneOf:
        - $ref: '#/components/schemas/DataQualityMonitor'
        - $ref: '#/components/schemas/PerformanceMonitor'
        - $ref: '#/components/schemas/DriftMonitor'
        - $ref: '#/components/schemas/CustomMetricMonitor'
        - $ref: '#/components/schemas/TracingMonitor'
      discriminator:
        propertyName: type
        mapping:
          DATA_QUALITY: '#/components/schemas/DataQualityMonitor'
          PERFORMANCE: '#/components/schemas/PerformanceMonitor'
          DRIFT: '#/components/schemas/DriftMonitor'
          CUSTOM_METRIC: '#/components/schemas/CustomMetricMonitor'
          TRACING: '#/components/schemas/TracingMonitor'
      description: >
        A monitor. The `type` field discriminates which variant (and therefore
        which type-specific fields) applies.
    Problem:
      type: object
      description: RFC 9457 Problem Details
      properties:
        title:
          type: string
          description: A short, human-readable summary of the problem type
        status:
          type: integer
          description: >-
            The HTTP status code generated by the origin server for this
            occurrence of the problem
        type:
          type: string
          format: uri-reference
          description: A URI reference that identifies the problem type
        detail:
          type: string
          description: >-
            A human-readable explanation specific to this occurrence of the
            problem
        instance:
          type: string
          format: uri-reference
          description: >-
            A URI reference that identifies the specific occurrence of the
            problem
      required:
        - title
        - status
      additionalProperties: false
      x-forward-compatible: true
    ForbiddenProblem:
      type: object
      description: >
        RFC 9457 Problem Details for a 403. When the account's plan doesn't
        include the capability,

        `entitlement` names it and `required_plan` says which plan unlocks it.
        When the account has

        reached its plan's limit on how many of something it may have,
        `entitlement` names the limit,

        `limit` and `current_count` describe it and `required_plan` says which
        plan raises it. When the

        caller lacks a permission, all of these are absent.
      properties:
        title:
          type: string
          description: A short, human-readable summary of the problem type
        status:
          type: integer
          description: >-
            The HTTP status code generated by the origin server for this
            occurrence of the problem
        type:
          type: string
          format: uri-reference
          description: A URI reference that identifies the problem type
        detail:
          type: string
          description: >-
            A human-readable explanation specific to this occurrence of the
            problem
        instance:
          type: string
          format: uri-reference
          description: >-
            A URI reference that identifies the specific occurrence of the
            problem
        entitlement:
          type: string
          description: >-
            Set when the plan doesn't include the capability or the account has
            reached its plan's limit on a count; the uppercase key, such as
            `DASHBOARD_CUSTOMIZATION` or `SPACES`.
        limit:
          type: integer
          description: >-
            Set when `entitlement` is a count limit; how many the account's plan
            allows.
        current_count:
          type: integer
          description: Set when `entitlement` is a count limit; how many the account has.
        required_plan:
          $ref: '#/components/schemas/UpgradePlan'
      required:
        - title
        - status
      additionalProperties: false
      x-forward-compatible: true
    CreateTracingMonitorRequest:
      type: object
      additionalProperties: false
      required:
        - type
        - project_id
        - name
        - metric
        - dimension
        - threshold
      properties:
        type:
          type: string
          enum:
            - TRACING
          description: Creates a monitor over a project's tracing data.
        project_id:
          type: string
          description: The LLM tracing project to monitor (base64 global ID).
        name:
          type: string
          minLength: 1
          maxLength: 255
          description: |
            Human-readable name of the monitor. 1 to 255 characters after
            surrounding whitespace is trimmed.
        metric:
          allOf:
            - $ref: '#/components/schemas/TracingDataQualityMetric'
          description: >-
            The metric computed over `dimension`. Must suit the attribute's data
            type.
        dimension:
          allOf:
            - $ref: '#/components/schemas/TracingAttributeInput'
          description: The span attribute or LLM eval the metric is computed over.
        threshold:
          $ref: '#/components/schemas/ThresholdConfigInput'
          description: The condition that triggers the monitor.
        evaluation_window_length_seconds:
          type: integer
          minimum: 300
          maximum: 31536000
          default: 86400
          description: >
            Length of the evaluated window, in seconds: 300, 600, 900, 1800, or
            a

            whole number of hours up to 8760 hours.
        delay_seconds:
          type: integer
          minimum: 0
          maximum: 15552000
          default: 0
          description: >
            Delay before a window is evaluated, in seconds, so late-arriving
            spans

            are counted. 0 to 15552000.
        filters:
          type: array
          maxItems: 25
          items:
            $ref: '#/components/schemas/TracingFilterInput'
          description: |
            Filters narrowing the spans the metric is computed over. At most 25.
            Defaults to no filters.
        notification_configs:
          type: array
          maxItems: 50
          items:
            $ref: '#/components/schemas/NotificationConfigInput'
          description: |
            Destinations notified on a triggered transition. At most 50, with no
            duplicate destination. Defaults to none.
        notifications_enabled:
          type: boolean
          default: true
          description: Whether notifications fire on a triggered transition.
        manual_evaluation_enabled:
          type: boolean
          default: false
          description: >
            Evaluate the monitor only on demand. Cannot be combined with an
            enabled

            `scheduled_runtime` or with `downtime`.
        downtime:
          allOf:
            - $ref: '#/components/schemas/DowntimeConfigInput'
          description: >-
            A recurring pause during which the monitor does not trigger.
            Defaults to none.
        scheduled_runtime:
          allOf:
            - $ref: '#/components/schemas/ScheduledRuntimeConfigInput'
          description: |
            When the monitor evaluates. Defaults to a disabled schedule, so the
            monitor evaluates continuously.
        notes:
          type: string
          maxLength: 1000
          description: >
            Notes included in notifications, at most 1000 characters. Supports
            the

            monitor notes variables; `{{threshold2.*}}` variables require a
            range

            threshold.
    CreateDataQualityMonitorRequest:
      type: object
      additionalProperties: false
      required:
        - type
        - project_id
        - name
        - metric
        - dimension
        - threshold
      properties:
        type:
          type: string
          enum:
            - DATA_QUALITY
          description: Creates a data quality monitor over an ML project's data.
        project_id:
          type: string
          description: The ML project to monitor (base64 global ID).
        name:
          type: string
          minLength: 1
          maxLength: 255
          description: |
            Human-readable name of the monitor. 1 to 255 characters after
            surrounding whitespace is trimmed.
        metric:
          allOf:
            - $ref: '#/components/schemas/DataQualityMetric'
          description: |
            The metric computed over `dimension`. Numeric dimensions support
            `COUNT`, `PERCENT_EMPTY`, `AVG`, `SUM`, `STANDARD_DEVIATION`, `P50`,
            `P95`, `P99` and `P99_9`. Categorical dimensions support `COUNT`,
            `PERCENT_EMPTY`, `CARDINALITY`, `NEW_VALUES` and `MISSING_VALUES`.
            String-list dimensions support `AVERAGE_STRING_LIST_LENGTH`, `COUNT`
            and `PERCENT_EMPTY`. `NEW_VALUES` and `MISSING_VALUES` require
            `baseline_config`.
        dimension:
          allOf:
            - $ref: '#/components/schemas/MlDimensionInput'
          description: >-
            The field the metric is computed over. Embedding features cannot be
            monitored.
        environment:
          allOf:
            - $ref: '#/components/schemas/MlMonitorEnvironment'
          default: PRODUCTION
          description: The environment whose data is monitored.
        model_versions:
          $ref: '#/components/schemas/MonitorModelVersionsInput'
        baseline_config:
          $ref: '#/components/schemas/BaselineConfigInput'
          description: |
            The comparison dataset. Required for `NEW_VALUES` and
            `MISSING_VALUES`, and rejected for every other metric.
        threshold:
          $ref: '#/components/schemas/ThresholdConfigInput'
          description: The condition that triggers the monitor.
        evaluation_window_length_seconds:
          type: integer
          minimum: 300
          maximum: 31536000
          default: 86400
          description: >
            Length of the evaluated window, in seconds: 300, 600, 900, 1800, or
            a

            whole number of hours up to 8760 hours.
        delay_seconds:
          type: integer
          minimum: 0
          maximum: 15552000
          default: 0
          description: >
            Delay before a window is evaluated, in seconds, so late-arriving
            data

            is counted. 0 to 15552000.
        filters:
          type: array
          maxItems: 25
          items:
            $ref: '#/components/schemas/MlFilterInput'
          description: |
            Filters narrowing the data the metric is computed over. At most 25.
            Defaults to no filters.
        notification_configs:
          type: array
          maxItems: 50
          items:
            $ref: '#/components/schemas/NotificationConfigInput'
          description: |
            Destinations notified on a triggered transition. At most 50, with no
            duplicate destination. Defaults to none.
        notifications_enabled:
          type: boolean
          default: true
          description: Whether notifications fire on a triggered transition.
        manual_evaluation_enabled:
          type: boolean
          default: false
          description: >
            Evaluate the monitor only on demand. Cannot be combined with an
            enabled

            `scheduled_runtime` or with `downtime`.
        downtime:
          allOf:
            - $ref: '#/components/schemas/DowntimeConfigInput'
          description: >-
            A recurring pause during which the monitor does not trigger.
            Defaults to none.
        scheduled_runtime:
          allOf:
            - $ref: '#/components/schemas/ScheduledRuntimeConfigInput'
          description: |
            When the monitor evaluates. Defaults to a disabled schedule, so the
            monitor evaluates continuously.
        notes:
          type: string
          maxLength: 1000
          description: >
            Notes included in notifications, at most 1000 characters. Supports
            the

            monitor notes variables; `{{threshold2.*}}` variables require a
            range

            threshold.
    DataQualityMonitor:
      x-forward-compatible: true
      allOf:
        - $ref: '#/components/schemas/MonitorBase'
        - type: object
          required:
            - metric
            - dimension
            - model_versions
          properties:
            type:
              type: string
              enum:
                - DATA_QUALITY
              description: Identifies this monitor as a data quality monitor.
            metric:
              $ref: '#/components/schemas/DataQualityMetric'
              description: The data quality metric this monitor evaluates.
            dimension:
              $ref: '#/components/schemas/Dimension'
              description: The field the metric is computed over.
            environment:
              $ref: '#/components/schemas/MlMonitorEnvironment'
              description: >-
                The model environment whose data the monitor evaluates. Omitted
                when the monitor reads an environment other than these.
            model_versions:
              type: array
              items:
                type: string
              description: >
                The model versions the metric is scoped to. An empty array means
                all model versions.
            baseline_config:
              $ref: '#/components/schemas/BaselineConfig'
              description: >
                The comparison dataset the metric is evaluated against. Omitted
                for monitors that evaluate the metric directly rather than by
                comparison.
    PerformanceMonitor:
      x-forward-compatible: true
      allOf:
        - $ref: '#/components/schemas/MonitorBase'
        - type: object
          required:
            - metric
            - model_versions
          properties:
            type:
              type: string
              enum:
                - PERFORMANCE
              description: Identifies this monitor as a performance monitor.
            metric:
              $ref: '#/components/schemas/PerformanceMetric'
              description: The model performance metric this monitor evaluates.
            environment:
              $ref: '#/components/schemas/MlMonitorEnvironment'
              description: >-
                The model environment whose data the monitor evaluates. Omitted
                when the monitor reads an environment other than these.
            model_versions:
              type: array
              items:
                type: string
              description: >
                The model versions the metric is scoped to. An empty array means
                all model versions.
            positive_class_value:
              type: string
              description: >-
                The positive class used for classification metrics. Omitted when
                unset.
            metric_at_ranking_k_value:
              type: integer
              description: >-
                The `k` cutoff for ranking metrics (e.g. NDCG@k). Omitted when
                unset.
    DriftMonitor:
      x-forward-compatible: true
      allOf:
        - $ref: '#/components/schemas/MonitorBase'
        - type: object
          required:
            - metric
            - dimension
            - baseline_config
            - model_versions
          properties:
            type:
              type: string
              enum:
                - DRIFT
              description: Identifies this monitor as a drift monitor.
            metric:
              $ref: '#/components/schemas/DriftMetric'
              description: The statistical drift metric this monitor evaluates.
            dimension:
              $ref: '#/components/schemas/Dimension'
              description: The feature or output the drift metric is computed over.
            environment:
              $ref: '#/components/schemas/MlMonitorEnvironment'
              description: >-
                The model environment whose data the monitor evaluates. Omitted
                when the monitor reads an environment other than these.
            baseline_config:
              $ref: '#/components/schemas/BaselineConfig'
              description: The comparison dataset the drift metric is evaluated against.
            model_versions:
              type: array
              items:
                type: string
              description: >
                The model versions the metric is scoped to. An empty array means
                all model versions.
    CustomMetricMonitor:
      x-forward-compatible: true
      allOf:
        - $ref: '#/components/schemas/MonitorBase'
        - type: object
          required:
            - custom_metric_id
            - model_versions
          properties:
            type:
              type: string
              enum:
                - CUSTOM_METRIC
              description: Identifies this monitor as a custom metric monitor.
            custom_metric_id:
              type: string
              description: |
                The custom metric this monitor evaluates (base64 global ID). 
            environment:
              $ref: '#/components/schemas/MlMonitorEnvironment'
              description: >-
                The model environment whose data the monitor evaluates. Omitted
                when the monitor reads an environment other than these.
            model_versions:
              type: array
              items:
                type: string
              description: >
                The model versions the metric is scoped to. An empty array means
                all model versions.
    TracingMonitor:
      x-forward-compatible: true
      allOf:
        - $ref: '#/components/schemas/MonitorBase'
        - type: object
          required:
            - metric
            - dimension
          properties:
            type:
              type: string
              enum:
                - TRACING
              description: Identifies this monitor as a tracing monitor.
            metric:
              allOf:
                - $ref: '#/components/schemas/TracingDataQualityMetric'
              description: Tracing monitors support only these data quality metrics.
            dimension:
              allOf:
                - $ref: '#/components/schemas/Dimension'
              description: >-
                The span/trace-derived field the metric is computed over.
                **Restricted to span_attribute, llm_eval, and user_annotation
                only.** 
    UpgradePlan:
      type: string
      description: >-
        The lowest plan that includes the capability. `PRO` can be purchased in
        the product; `ENTERPRISE` requires contacting Arize sales.
      enum:
        - PRO
        - ENTERPRISE
    TracingDataQualityMetric:
      type: string
      enum:
        - COUNT
        - PERCENT_EMPTY
        - CARDINALITY
        - AVG
        - SUM
        - STANDARD_DEVIATION
        - P50
        - P95
        - P99
        - P99_9
      description: >
        The data quality metrics computed by a tracing monitor.

        Numeric span attributes support aggregations such as `SUM` and `AVG`;

        categorical attributes support `COUNT`, `PERCENT_EMPTY`, and
        `CARDINALITY`.

        Comparison metrics (`NEW_VALUES`, `MISSING_VALUES`) are not supported.
    TracingAttributeInput:
      type: object
      additionalProperties: false
      required:
        - category
        - name
      properties:
        category:
          $ref: '#/components/schemas/TracingAttributeCategory'
        name:
          type: string
          minLength: 1
          description: |
            Exact, case-sensitive name of the attribute as it appears in the
            project's recent tracing data. LLM eval names include their scope
            prefix, for example `eval.correctness.score`.
    ThresholdConfigInput:
      oneOf:
        - $ref: '#/components/schemas/ManualSingleThresholdInput'
        - $ref: '#/components/schemas/DynamicSingleThresholdInput'
        - $ref: '#/components/schemas/ManualRangeThresholdInput'
        - $ref: '#/components/schemas/DynamicRangeThresholdInput'
      discriminator:
        propertyName: type
        mapping:
          MANUAL_SINGLE: '#/components/schemas/ManualSingleThresholdInput'
          DYNAMIC_SINGLE: '#/components/schemas/DynamicSingleThresholdInput'
          MANUAL_RANGE: '#/components/schemas/ManualRangeThresholdInput'
          DYNAMIC_RANGE: '#/components/schemas/DynamicRangeThresholdInput'
      description: |
        The monitor's threshold, in the same shape a monitor returns. The `type`
        field selects a manual or dynamic threshold, single or a bounded range.
    TracingFilterInput:
      type: object
      additionalProperties: false
      required:
        - dimension
        - operator
        - values
      properties:
        dimension:
          allOf:
            - $ref: '#/components/schemas/TracingAttributeInput'
          description: The attribute the filter applies to.
        operator:
          allOf:
            - $ref: '#/components/schemas/MonitorFilterInputOperator'
          description: |
            Comparison applied to the attribute. Numeric attributes support
            `EQUALS`, `NOT_EQUALS`, `GREATER_THAN`, `GREATER_THAN_OR_EQUAL`,
            `LESS_THAN` and `LESS_THAN_OR_EQUAL`; string attributes support
            `EQUALS`, `NOT_EQUALS` and `CONTAINS_STRING`; string-list attributes
            support `CONTAINS`.
        values:
          type: array
          minItems: 1
          maxItems: 100
          items:
            type: string
            minLength: 1
            maxLength: 1000
          description: |
            Values compared against, at most 100, each 1 to 1000 characters.
            Numeric attributes take decimal strings such as `"0.5"`. The
            `GREATER_THAN`, `GREATER_THAN_OR_EQUAL`, `LESS_THAN` and
            `LESS_THAN_OR_EQUAL` operators take exactly one value.
    NotificationConfigInput:
      oneOf:
        - $ref: '#/components/schemas/EmailNotificationConfigInput'
        - $ref: '#/components/schemas/IntegrationNotificationConfigInput'
        - $ref: '#/components/schemas/WebhookNotificationConfigInput'
      discriminator:
        propertyName: type
        mapping:
          EMAIL: '#/components/schemas/EmailNotificationConfigInput'
          INTEGRATION: '#/components/schemas/IntegrationNotificationConfigInput'
          WEBHOOK: '#/components/schemas/WebhookNotificationConfigInput'
      description: >-
        A destination notified on a triggered transition. `type` selects the
        channel.
    DowntimeConfigInput:
      type: object
      additionalProperties: false
      required:
        - start
        - duration_seconds
        - frequency_days
      properties:
        start:
          type: string
          format: date-time
          description: >-
            When the first downtime window begins, as an RFC 3339 timestamp with
            an offset.
        duration_seconds:
          type: integer
          description: |
            Length of each downtime window, in seconds. A whole number of hours
            from 1 to 1344, and no longer than `frequency_days`.
        frequency_days:
          type: integer
          minimum: 1
          maximum: 56
          description: >-
            Days between the starts of successive downtime windows, from 1 to
            56.
    ScheduledRuntimeConfigInput:
      type: object
      additionalProperties: false
      required:
        - enabled
      properties:
        enabled:
          type: boolean
          description: Whether the monitor evaluates only on the schedule below.
        cadence_seconds:
          type: integer
          description: >
            Interval between evaluations, in seconds: 300, 600, 900, 1800, 1 to
            24

            hours, 1 to 30 days, 1 to 4 weeks, or 1 to 12 months of 30 days.

            Required when `enabled` is true; defaults to 10800 otherwise.
        days_of_week:
          type: array
          minItems: 1
          maxItems: 7
          uniqueItems: true
          items:
            type: integer
            minimum: 0
            maximum: 6
          description: |
            Days the monitor may evaluate on, from `0` (Sunday) to `6`
            (Saturday). Required when `enabled` is true. Defaults to every day
            when omitted.
    DataQualityMetric:
      type: string
      enum:
        - COUNT
        - PERCENT_EMPTY
        - CARDINALITY
        - NEW_VALUES
        - MISSING_VALUES
        - AVG
        - SUM
        - STANDARD_DEVIATION
        - P50
        - P95
        - P99
        - P99_9
        - AVERAGE_STRING_LIST_LENGTH
      description: |
        The data quality metric computed over the selected dimension.
    MlDimensionInput:
      type: object
      additionalProperties: false
      required:
        - category
        - name
      properties:
        category:
          $ref: '#/components/schemas/MlDimensionCategory'
        name:
          type: string
          minLength: 1
          description: |
            Exact, case-sensitive name of the feature, tag, prediction or actual
            in the project's schema for the monitored environment, for example
            `fico_score` or `scorePredictionLabel`.
    MlMonitorEnvironment:
      type: string
      enum:
        - PRODUCTION
        - TRAINING
        - VALIDATION
      description: The model environment whose data a monitor evaluates.
    MonitorModelVersionsInput:
      type: array
      maxItems: 100
      uniqueItems: true
      items:
        type: string
        minLength: 1
      description: |
        External model version IDs the data is limited to, at most 100 with no
        duplicates, each an existing version of the project. Defaults to all
        versions.
    BaselineConfigInput:
      oneOf:
        - $ref: '#/components/schemas/ModelBaselineConfigInput'
        - $ref: '#/components/schemas/CustomBaselineConfigInput'
      discriminator:
        propertyName: type
        mapping:
          MODEL_BASELINE: '#/components/schemas/ModelBaselineConfigInput'
          CUSTOM_BASELINE: '#/components/schemas/CustomBaselineConfigInput'
      description: |
        The comparison dataset a metric is evaluated against. `type` selects the
        project's primary baseline or a custom production window.
    MlFilterInput:
      type: object
      additionalProperties: false
      required:
        - dimension
        - operator
        - values
      properties:
        dimension:
          allOf:
            - $ref: '#/components/schemas/MlDimensionInput'
          description: >-
            The field the filter applies to. Embedding features cannot be
            filtered.
        operator:
          allOf:
            - $ref: '#/components/schemas/MonitorFilterInputOperator'
          description: >
            Comparison applied to the field. Numeric fields support `EQUALS`,

            `NOT_EQUALS`, `GREATER_THAN`, `GREATER_THAN_OR_EQUAL`, `LESS_THAN`
            and

            `LESS_THAN_OR_EQUAL`; categorical fields support `EQUALS`,

            `NOT_EQUALS` and `CONTAINS_STRING`; string-list fields support

            `CONTAINS`.
        values:
          type: array
          minItems: 1
          maxItems: 100
          items:
            type: string
            minLength: 1
            maxLength: 1000
          description: |
            Values compared against, at most 100, each 1 to 1000 characters.
            Numeric fields take decimal strings such as `"0.5"`. The
            `GREATER_THAN`, `GREATER_THAN_OR_EQUAL`, `LESS_THAN` and
            `LESS_THAN_OR_EQUAL` operators take exactly one value.
    MonitorBase:
      type: object
      x-forward-compatible: true
      required:
        - id
        - name
        - type
        - project_id
        - uri
        - status
        - created_at
        - updated_at
        - threshold
        - notifications_enabled
        - manual_evaluation_enabled
        - created_by_user_id
      properties:
        id:
          type: string
          description: Unique identifier for the monitor (base64 global ID).
        name:
          type: string
          description: Human-readable name of the monitor.
        type:
          allOf:
            - $ref: '#/components/schemas/MonitorType'
          description: Required on create. Immutable after creation.
        project_id:
          type: string
          description: |
            The project that the monitor belongs to (base64 global ID).
        uri:
          type: string
          readOnly: true
          description: The UI deep link to the monitor.
        status:
          allOf:
            - $ref: '#/components/schemas/MonitorStatus'
          description: Current evaluation state. Read-only.
        threshold:
          $ref: '#/components/schemas/ThresholdConfig'
          description: The condition that determines when this monitor triggers.
        notifications_enabled:
          type: boolean
          default: true
          description: Whether notifications fire on a triggered transition.
        manual_evaluation_enabled:
          type: boolean
          default: false
          description: >-
            Whether the monitor is evaluated manually rather than on the
            automatic cadence.
        created_by_user_id:
          type: string
          description: The user who created the monitor (base64 global ID).
        filters:
          type: array
          items:
            $ref: '#/components/schemas/MonitorFilter'
          description: >
            Data filters applied to the monitor's metric. Omitted when the
            metric is computed over all data.
        notification_configs:
          type: array
          items:
            $ref: '#/components/schemas/NotificationConfig'
          description: >
            Notification channels (email / integration / webhook) notified on a
            triggered transition. Omitted when no channels are configured.
        created_at:
          type: string
          format: date-time
          readOnly: true
          description: When the monitor was created.
        updated_at:
          type: string
          format: date-time
          readOnly: true
          description: When the monitor was last updated.
        downtime:
          $ref: '#/components/schemas/DowntimeConfig'
          description: >
            A recurring period during which the monitor is paused and will not
            trigger. Omitted when no downtime is configured.
        scheduled_runtime:
          $ref: '#/components/schemas/ScheduledRuntimeConfig'
          description: |
            Returned when schedule configuration exists.
        evaluation_window_length_seconds:
          type: integer
          description: |
            The length of the evaluation window in seconds. Omitted when unset.
        delay_seconds:
          type: integer
          description: >
            The delay applied before evaluating a window, in seconds. Omitted
            when unset.
        evaluated_at:
          type: string
          format: date-time
          readOnly: true
          description: >-
            The last time the metric was computed. Omitted if it has never been
            evaluated.
        latest_computed_value:
          type: number
          readOnly: true
          description: >-
            The most recently computed metric value. Omitted if it has never
            been evaluated.
        notes:
          type: string
          description: Free-form notes attached to the monitor. Omitted when unset.
    Dimension:
      type: object
      x-forward-compatible: true
      required:
        - category
      properties:
        category:
          allOf:
            - $ref: '#/components/schemas/DimensionCategory'
          description: The category of the monitored dimension.
        name:
          type: string
          description: >
            Name of the monitored field. Omitted when the category has no
            concrete field name. 
    BaselineConfig:
      oneOf:
        - $ref: '#/components/schemas/ModelBaselineConfig'
        - $ref: '#/components/schemas/CustomBaselineConfig'
      discriminator:
        propertyName: type
        mapping:
          MODEL_BASELINE: '#/components/schemas/ModelBaselineConfig'
          CUSTOM_BASELINE: '#/components/schemas/CustomBaselineConfig'
      description: >-
        The comparison dataset configuration used by drift and comparison-based
        data quality monitors. The `type` field determines whether the
        comparison dataset uses the model's primary baseline or a custom fixed
        or moving window.
    PerformanceMetric:
      type: string
      enum:
        - ACCURACY
        - CALIBRATION
        - F_1
        - FALSE_NEGATIVE_DENSITY
        - FALSE_NEGATIVE_RATE
        - FALSE_POSITIVE_RATE
        - PRECISION
        - RECALL
        - SENSITIVITY
        - SPECIFICITY
        - MICRO_AVERAGED_PRECISION
        - MACRO_AVERAGED_PRECISION
        - MULTI_CLASS_PRECISION
        - MICRO_AVERAGED_RECALL
        - MACRO_AVERAGED_RECALL
        - MULTI_CLASS_RECALL
        - ACTUALS_AVERAGE
        - MAE
        - MAPE
        - MEAN_ERROR
        - MSE
        - PREDICTION_AVERAGE
        - RMSE
        - R_SQUARED
        - SMAPE
        - WAPE
        - MASE
        - AUC
        - LOG_LOSS
        - NDCG
        - PR_AUC
        - GROUP_AUC
        - MRR
        - RANKING_MAP
        - RANKING_RECALL
        - RANKING_PRECISION
        - DISPARATE_IMPACT
        - FALSE_POSITIVE_RATE_PARITY
        - RECALL_PARITY
      description: |
        The model performance metric.
    DriftMetric:
      type: string
      enum:
        - PSI
        - KL
        - JS
        - KS
        - EUCLIDEAN_DISTANCE
        - COSINE_SIMILARITY
      description: >
        The statistical drift metric. `PSI`/`KL`/`JS`/`KS` apply to structured
        data;

        `EUCLIDEAN_DISTANCE`/`COSINE_SIMILARITY` apply to unstructured
        (embedding) data.
    TracingAttributeCategory:
      type: string
      enum:
        - SPAN_ATTRIBUTE
        - LLM_EVAL
      description: >-
        The kind of tracing attribute a tracing monitor or filter is computed
        over.
    ManualSingleThresholdInput:
      type: object
      additionalProperties: false
      required:
        - type
        - operator
        - value
      properties:
        type:
          type: string
          enum:
            - MANUAL_SINGLE
          description: Identifies this threshold as a single, manually-set value.
        operator:
          allOf:
            - $ref: '#/components/schemas/ThresholdOperator'
          description: >-
            Comparison between the computed metric and `value`. Any operator
            except `NOT_EQUALS`.
        value:
          type: number
          description: The value the computed metric is compared against.
    DynamicSingleThresholdInput:
      type: object
      additionalProperties: false
      required:
        - type
        - calculation
        - operator
        - multiplier
      properties:
        type:
          type: string
          enum:
            - DYNAMIC_SINGLE
          description: >-
            Identifies this threshold as a single value derived from the
            metric's history.
        calculation:
          allOf:
            - $ref: '#/components/schemas/ThresholdCalculation'
          description: The statistical calculation the threshold is derived from.
        operator:
          allOf:
            - $ref: '#/components/schemas/ThresholdOperator'
          description: >-
            Comparison between the computed metric and the derived threshold.
            Any operator except `NOT_EQUALS`.
        multiplier:
          type: number
          minimum: 0
          exclusiveMinimum: true
          maximum: 10
          description: >-
            Number of deviations from the baseline. Greater than 0 and at most
            10.
    ManualRangeThresholdInput:
      type: object
      additionalProperties: false
      required:
        - type
        - lower
        - upper
      properties:
        type:
          type: string
          enum:
            - MANUAL_RANGE
          description: Identifies this threshold as a manually-set lower/upper bound range.
        lower:
          allOf:
            - $ref: '#/components/schemas/ManualThresholdBoundInput'
          description: |
            The lower bound. Its operator must be `GREATER_THAN` or
            `GREATER_THAN_OR_EQUAL`, and its value must be below `upper.value`.
        upper:
          allOf:
            - $ref: '#/components/schemas/ManualThresholdBoundInput'
          description: >-
            The upper bound. Its operator must be `LESS_THAN` or
            `LESS_THAN_OR_EQUAL`.
    DynamicRangeThresholdInput:
      type: object
      additionalProperties: false
      required:
        - type
        - calculation
        - lower
        - upper
      properties:
        type:
          type: string
          enum:
            - DYNAMIC_RANGE
          description: >-
            Identifies this threshold as a lower/upper bound range derived from
            the metric's history.
        calculation:
          allOf:
            - $ref: '#/components/schemas/ThresholdCalculation'
          description: The statistical calculation both bounds are derived from.
        lower:
          allOf:
            - $ref: '#/components/schemas/DynamicThresholdBoundInput'
          description: >-
            The lower bound. Its operator must be `GREATER_THAN` or
            `GREATER_THAN_OR_EQUAL`.
        upper:
          allOf:
            - $ref: '#/components/schemas/DynamicThresholdBoundInput'
          description: >-
            The upper bound. Its operator must be `LESS_THAN` or
            `LESS_THAN_OR_EQUAL`.
    MonitorFilterInputOperator:
      type: string
      enum:
        - EQUALS
        - NOT_EQUALS
        - GREATER_THAN
        - GREATER_THAN_OR_EQUAL
        - LESS_THAN
        - LESS_THAN_OR_EQUAL
        - CONTAINS
        - CONTAINS_STRING
      description: Operators a monitor's filter can use.
    EmailNotificationConfigInput:
      type: object
      additionalProperties: false
      required:
        - type
        - email_address
      properties:
        type:
          type: string
          enum:
            - EMAIL
          description: Identifies this notification channel as an email address.
        email_address:
          type: string
          format: email
          maxLength: 254
          description: >-
            A valid email address, at most 254 characters, notified on a
            triggered transition.
    IntegrationNotificationConfigInput:
      type: object
      additionalProperties: false
      required:
        - type
        - integration_id
      properties:
        type:
          type: string
          enum:
            - INTEGRATION
          description: >-
            Identifies this notification channel as a connected alert
            integration.
        integration_id:
          type: string
          description: >-
            The alert integration to notify (base64 global ID). It must belong
            to the project's organization.
    WebhookNotificationConfigInput:
      type: object
      additionalProperties: false
      required:
        - type
        - id
      properties:
        type:
          type: string
          enum:
            - WEBHOOK
          description: Identifies this notification channel as a webhook.
        id:
          type: string
          description: |
            The webhook to notify (base64 global ID). It must belong to the
            project's organization and is subscribed to every monitor event.
    MlDimensionCategory:
      type: string
      enum:
        - FEATURE_LABEL
        - TAG
        - PREDICTIONS
        - PREDICTIONS_SCORE
        - PREDICTIONS_CLASS
        - ACTUALS
        - ACTUALS_SCORE
        - ACTUALS_CLASS
      description: >
        The kind of ML project field a monitor or filter is computed over. Model

        versions are selected with `model_versions`, not as a dimension or
        filter.
    ModelBaselineConfigInput:
      type: object
      additionalProperties: false
      required:
        - type
      properties:
        type:
          type: string
          enum:
            - MODEL_BASELINE
          description: |
            Compare against the project's primary baseline. Rejected when the
            project has no primary baseline.
    CustomBaselineConfigInput:
      type: object
      additionalProperties: false
      required:
        - type
        - window
      properties:
        type:
          type: string
          enum:
            - CUSTOM_BASELINE
          description: Compare against a fixed or moving window of production data.
        window:
          $ref: '#/components/schemas/CustomBaselineWindowInput'
          description: The production date range or rolling window compared against.
        model_versions:
          $ref: '#/components/schemas/MonitorModelVersionsInput'
        filters:
          type: array
          maxItems: 25
          items:
            $ref: '#/components/schemas/MlFilterInput'
          description: >-
            Filters narrowing the comparison dataset, on fields in the project's
            production data. At most 25. Defaults to no filters.
    MonitorType:
      type: string
      enum:
        - DATA_QUALITY
        - PERFORMANCE
        - DRIFT
        - CUSTOM_METRIC
        - TRACING
      description: >
        The kind of monitor. Determines which type-specific fields apply.

        - `DATA_QUALITY` - Monitors a data quality metric (e.g. percent empty,
        CARDINALITY).

        - `PERFORMANCE` - Monitors a model performance metric (e.g. ACCURACY,
        RMSE).

        - `DRIFT` - Monitors distributional drift of a feature/output.

        - `CUSTOM_METRIC` - Monitors a user-defined custom metric.

        - `TRACING` - Monitors a span/trace-derived metric (e.g. span
        attributes, evals).
    MonitorStatus:
      type: string
      enum:
        - TRIGGERED
        - CLEARED
        - NO_DATA
      description: |
        The monitor's current state from its most recent evaluation.
        - `TRIGGERED` - The metric breached the threshold.
        - `CLEARED` - The metric is within healthy bounds.
        - `NO_DATA` - No data was available to evaluate.
    ThresholdConfig:
      oneOf:
        - $ref: '#/components/schemas/ManualSingleThreshold'
        - $ref: '#/components/schemas/DynamicSingleThreshold'
        - $ref: '#/components/schemas/ManualRangeThreshold'
        - $ref: '#/components/schemas/DynamicRangeThreshold'
      discriminator:
        propertyName: type
        mapping:
          MANUAL_SINGLE: '#/components/schemas/ManualSingleThreshold'
          DYNAMIC_SINGLE: '#/components/schemas/DynamicSingleThreshold'
          MANUAL_RANGE: '#/components/schemas/ManualRangeThreshold'
          DYNAMIC_RANGE: '#/components/schemas/DynamicRangeThreshold'
      description: >
        The monitor's threshold. The `type` field discriminates whether the
        threshold is manual or dynamic, and single or a bounded range.
    MonitorFilter:
      type: object
      x-forward-compatible: true
      required:
        - dimension
        - operator
        - values
      properties:
        dimension:
          $ref: '#/components/schemas/Dimension'
          description: The field the filter is applied to.
        operator:
          $ref: '#/components/schemas/FilterOperator'
          description: >-
            The comparison applied between the field's value and the filter's
            values.
        values:
          type: array
          items:
            type: string
          description: The values compared against by the operator.
    NotificationConfig:
      oneOf:
        - $ref: '#/components/schemas/EmailNotificationConfig'
        - $ref: '#/components/schemas/IntegrationNotificationConfig'
        - $ref: '#/components/schemas/WebhookNotificationConfig'
      discriminator:
        propertyName: type
        mapping:
          EMAIL: '#/components/schemas/EmailNotificationConfig'
          INTEGRATION: '#/components/schemas/IntegrationNotificationConfig'
          WEBHOOK: '#/components/schemas/WebhookNotificationConfig'
    DowntimeConfig:
      type: object
      x-forward-compatible: true
      required:
        - start
        - duration_seconds
        - frequency_days
      properties:
        start:
          type: string
          format: date-time
          description: When the downtime window begins.
        duration_seconds:
          type: integer
          description: How long each downtime window lasts, in seconds.
        frequency_days:
          type: integer
          description: How often the downtime window repeats, in days.
    ScheduledRuntimeConfig:
      type: object
      x-forward-compatible: true
      required:
        - enabled
      properties:
        enabled:
          type: boolean
          description: >-
            Whether the monitor runs on a schedule. `false` means automatic
            scheduled evaluation is disabled.
        cadence_seconds:
          type: integer
          description: How often the monitor evaluates, in seconds.
        days_of_week:
          type: array
          items:
            type: integer
            minimum: 0
            maximum: 6
          description: >
            Days of the week the monitor runs on (`0` = Sunday … `6` =
            Saturday).
    DimensionCategory:
      type: string
      enum:
        - FEATURE_LABEL
        - TAG
        - MODEL_VERSION
        - BATCH_ID
        - SPAN_ATTRIBUTE
        - LLM_EVAL
        - USER_ANNOTATION
        - ACTUALS
        - ACTUALS_SCORE
        - ACTUALS_CLASS
        - PREDICTIONS
        - PREDICTIONS_SCORE
        - PREDICTIONS_CLASS
      description: |
        The category of dimension the metric is evaluated over.
    ModelBaselineConfig:
      type: object
      x-forward-compatible: true
      required:
        - type
        - filters
      properties:
        type:
          type: string
          enum:
            - MODEL_BASELINE
          description: Identifies this comparison dataset as the model's primary baseline.
        filters:
          type: array
          items:
            $ref: '#/components/schemas/MonitorFilter'
          description: >
            Filters applied to the comparison dataset. An empty array means no
            comparison dataset filters are configured.
      description: |
        Uses the model's primary baseline as the comparison dataset.
    CustomBaselineConfig:
      type: object
      x-forward-compatible: true
      required:
        - type
        - filters
        - model_versions
        - window
      properties:
        type:
          type: string
          enum:
            - CUSTOM_BASELINE
          description: >-
            Identifies this comparison dataset as a custom fixed or moving
            window.
        filters:
          type: array
          items:
            $ref: '#/components/schemas/MonitorFilter'
          description: >
            Filters applied to the comparison dataset. An empty array means no
            comparison dataset filters are configured.
        model_versions:
          type: array
          items:
            type: string
          description: >
            Model versions included in the comparison dataset. An empty array
            means all model versions.
        window:
          $ref: '#/components/schemas/CustomBaselineWindow'
          description: >-
            The custom date range or rolling window used as the comparison
            dataset.
      description: |
        Uses a custom fixed or moving window as the comparison dataset.
    ThresholdOperator:
      type: string
      enum:
        - GREATER_THAN
        - GREATER_THAN_OR_EQUAL
        - LESS_THAN
        - LESS_THAN_OR_EQUAL
        - EQUALS
        - NOT_EQUALS
      description: >
        Numeric comparison operators used when evaluating a computed metric
        value against a threshold.
    ThresholdCalculation:
      type: string
      enum:
        - STDEV
        - MAD
      description: |
        The statistical calculation used to derive a dynamic threshold.
        - `STDEV` - standard deviation of the metric over the baseline.
        - `MAD` - median absolute deviation of the metric over the baseline.
    ManualThresholdBoundInput:
      type: object
      additionalProperties: false
      required:
        - operator
        - value
      properties:
        operator:
          allOf:
            - $ref: '#/components/schemas/ThresholdOperator'
          description: Comparison between the computed metric and this bound.
        value:
          type: number
          description: The bound value.
    DynamicThresholdBoundInput:
      type: object
      additionalProperties: false
      required:
        - operator
        - multiplier
      properties:
        operator:
          allOf:
            - $ref: '#/components/schemas/ThresholdOperator'
          description: Comparison between the computed metric and this bound.
        multiplier:
          type: number
          minimum: 0
          exclusiveMinimum: true
          maximum: 10
          description: >-
            Number of deviations from the baseline for this bound. Greater than
            0 and at most 10.
    CustomBaselineWindowInput:
      oneOf:
        - $ref: '#/components/schemas/FixedCustomBaselineWindowInput'
        - $ref: '#/components/schemas/MovingCustomBaselineWindowInput'
      discriminator:
        propertyName: type
        mapping:
          FIXED: '#/components/schemas/FixedCustomBaselineWindowInput'
          MOVING: '#/components/schemas/MovingCustomBaselineWindowInput'
      description: >-
        The custom comparison window. `type` selects a fixed range or a moving
        window.
    ManualSingleThreshold:
      type: object
      x-forward-compatible: true
      required:
        - type
        - operator
        - value
      properties:
        type:
          type: string
          enum:
            - MANUAL_SINGLE
          description: Identifies this threshold as a single, manually-set value.
        operator:
          $ref: '#/components/schemas/ThresholdOperator'
          description: >-
            The comparison applied between the computed metric value and the
            threshold value.
        value:
          type: number
          description: The threshold value the computed metric is compared against.
    DynamicSingleThreshold:
      type: object
      x-forward-compatible: true
      required:
        - type
        - calculation
        - operator
        - multiplier
      properties:
        type:
          type: string
          enum:
            - DYNAMIC_SINGLE
          description: >-
            Identifies this threshold as a single value derived from a
            statistical calculation.
        calculation:
          $ref: '#/components/schemas/ThresholdCalculation'
          description: The statistical calculation used to derive the threshold value.
        operator:
          $ref: '#/components/schemas/ThresholdOperator'
          description: >-
            The comparison applied between the computed metric value and the
            derived threshold.
        multiplier:
          type: number
          description: |
            The multiplier applied to the calculation (e.g. number of standard
            deviations) to derive the threshold.
    ManualRangeThreshold:
      type: object
      x-forward-compatible: true
      required:
        - type
        - lower
        - upper
      properties:
        type:
          type: string
          enum:
            - MANUAL_RANGE
          description: Identifies this threshold as a manually-set lower/upper bound range.
        lower:
          allOf:
            - $ref: '#/components/schemas/ManualThresholdBound'
          description: >
            The lower bound of the range. Its operator must be `GREATER_THAN` or
            `GREATER_THAN_OR_EQUAL`.
        upper:
          allOf:
            - $ref: '#/components/schemas/ManualThresholdBound'
          description: >
            The upper bound of the range. Its `operator` must be `LESS_THAN` or
            `LESS_THAN_OR_EQUAL`.
    DynamicRangeThreshold:
      type: object
      x-forward-compatible: true
      required:
        - type
        - calculation
        - lower
        - upper
      properties:
        type:
          type: string
          enum:
            - DYNAMIC_RANGE
          description: >-
            Identifies this threshold as a lower/upper bound range derived from
            a statistical calculation.
        calculation:
          $ref: '#/components/schemas/ThresholdCalculation'
          description: >-
            The statistical calculation used to derive the lower and upper
            bounds.
        lower:
          allOf:
            - $ref: '#/components/schemas/DynamicThresholdBound'
          description: >
            The lower bound of the range. Its `operator` must be `GREATER_THAN`
            or `GREATER_THAN_OR_EQUAL`.
        upper:
          allOf:
            - $ref: '#/components/schemas/DynamicThresholdBound'
          description: >
            The upper bound of the range. Its `operator` must be `LESS_THAN` or
            `LESS_THAN_OR_EQUAL`.
    FilterOperator:
      type: string
      enum:
        - GREATER_THAN
        - GREATER_THAN_OR_EQUAL
        - LESS_THAN
        - LESS_THAN_OR_EQUAL
        - EQUALS
        - NOT_EQUALS
        - TOP_N
        - CONTAINS
        - CONTAINS_STRING
        - SIMILAR_TO
      description: >
        Operators used by a monitor's data filters. Includes string operators as
        well as numeric ones.
    EmailNotificationConfig:
      type: object
      x-forward-compatible: true
      required:
        - type
        - email_address
      properties:
        type:
          type: string
          enum:
            - EMAIL
          description: Identifies this notification channel as an email address.
        email_address:
          type: string
          format: email
          description: Email address notified on a triggered transition.
    IntegrationNotificationConfig:
      type: object
      x-forward-compatible: true
      required:
        - type
        - integration_id
      properties:
        type:
          type: string
          enum:
            - INTEGRATION
          description: Identifies this notification channel as a connected integration.
        integration_id:
          type: string
          description: The integration to notify (base64 global ID).
    WebhookNotificationConfig:
      type: object
      x-forward-compatible: true
      required:
        - type
        - id
      properties:
        type:
          type: string
          enum:
            - WEBHOOK
          description: Identifies this notification channel as a webhook.
        id:
          type: string
          description: The webhook to notify (base64 global ID).
        url:
          type: string
          nullable: true
          description: The webhook endpoint URL.
    CustomBaselineWindow:
      oneOf:
        - $ref: '#/components/schemas/FixedCustomBaselineWindow'
        - $ref: '#/components/schemas/MovingCustomBaselineWindow'
      discriminator:
        propertyName: type
        mapping:
          FIXED: '#/components/schemas/FixedCustomBaselineWindow'
          MOVING: '#/components/schemas/MovingCustomBaselineWindow'
      description: >
        The custom comparison window. The `type` field determines whether the
        window is fixed or moving.
    FixedCustomBaselineWindowInput:
      type: object
      additionalProperties: false
      required:
        - type
        - fixed_start_date
        - fixed_end_date
      properties:
        type:
          type: string
          enum:
            - FIXED
          description: Compare against data between two timestamps.
        fixed_start_date:
          type: string
          format: date-time
          description: >-
            Start of the comparison range, with a UTC offset. Must be before
            `fixed_end_date`.
        fixed_end_date:
          type: string
          format: date-time
          description: End of the comparison range, with a UTC offset.
    MovingCustomBaselineWindowInput:
      type: object
      additionalProperties: false
      required:
        - type
        - moving_window_seconds
        - moving_window_delay_seconds
      properties:
        type:
          type: string
          enum:
            - MOVING
          description: Compare against a window that moves with the evaluation time.
        moving_window_seconds:
          type: integer
          minimum: 3600
          maximum: 70956000
          description: >-
            Length of the moving window, in seconds. A whole number of hours
            from 1 to 19710.
        moving_window_delay_seconds:
          type: integer
          minimum: 0
          maximum: 70956000
          description: >-
            How far before the evaluation time the moving window ends, in
            seconds. A whole number of hours from 0 to 19710.
    ManualThresholdBound:
      type: object
      x-forward-compatible: true
      required:
        - operator
        - value
      properties:
        operator:
          $ref: '#/components/schemas/ThresholdOperator'
          description: >-
            The comparison applied between the computed metric value and this
            bound.
        value:
          type: number
          description: The bound value the computed metric is compared against.
    DynamicThresholdBound:
      type: object
      x-forward-compatible: true
      required:
        - operator
        - multiplier
      properties:
        operator:
          $ref: '#/components/schemas/ThresholdOperator'
          description: >-
            The comparison applied between the computed metric value and this
            bound.
        multiplier:
          type: number
          description: >
            The multiplier applied to the calculation (e.g. number of standard
            deviations) to derive this bound.
    FixedCustomBaselineWindow:
      type: object
      x-forward-compatible: true
      required:
        - type
        - fixed_start_date
        - fixed_end_date
      properties:
        type:
          type: string
          enum:
            - FIXED
          description: Identifies this comparison window as a fixed date range.
        fixed_start_date:
          type: string
          format: date-time
          description: The start of the fixed comparison window.
        fixed_end_date:
          type: string
          format: date-time
          description: The end of the fixed comparison window.
      description: >-
        A custom comparison dataset using data between a fixed start and end
        date.
    MovingCustomBaselineWindow:
      type: object
      x-forward-compatible: true
      required:
        - type
        - moving_window_seconds
        - moving_window_delay_seconds
      properties:
        type:
          type: string
          enum:
            - MOVING
          description: >-
            Identifies this comparison window as a moving window relative to
            now.
        moving_window_seconds:
          type: integer
          description: The length of the moving comparison window, in seconds.
        moving_window_delay_seconds:
          type: integer
          description: The delay before the moving comparison window, in seconds.
      description: A custom comparison dataset using a moving window defined in seconds.
  securitySchemes:
    bearerAuth:
      type: http
      scheme: bearer
      bearerFormat: <api-key>
      description: >
        Most Arize AI endpoints require authentication. For those endpoints that
        require authentication, include your API key in the request header using
        the format

        ``` Authorization: Bearer <api-key>

        ```

        Service accounts may instead present an access token issued by an
        external identity provider (Microsoft Entra ID) when the account admin
        has configured that provider. Send it in the same header:
        `Authorization: Bearer <access-token>`.

````