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

# Object graph

> Every field on Viewer, Account, AccountOrganization, Space and Model, the five types you traverse from the root queries to reach the rest of the GraphQL API.

The Arize GraphQL schema is a graph you walk from the [root queries](/docs/ax/graphql-reference/queries/root-queries.md). This page lists the fields on the five types every traversal passes through. Connection fields (types ending in `Connection`) are paginated and usually require `first` or `last`; see [Resource limitations](/docs/ax/graphql-reference/overview/resource-limitations.md).

The schema has 653 object types in total. For any type not listed here, use the [API explorer](/docs/ax/graphql-reference/overview/api-explorer.md) or an introspection query.

```graphql Minimal traversal theme={"theme":{"light":"github-light-default","dark":"github-dark-default"}}
query FindSpacesAndModels {
  viewer {
    spaces(first: 20) {
      edges {
        node {
          id
          name
          models(first: 50) {
            edges {
              node {
                id
                name
                projectType
              }
            }
          }
        }
      }
    }
  }
}
```

## Viewer

The authenticated caller. `viewer.spaces` lists every space the API key can see and is usually the fastest way to find a space ID. The current viewer context

**id** `ID!`

The ID of an object

**user** `User!`

The user entity backing this viewer.

**spaces(after: String, first: Int, before: String, last: Int, search: String, sort: \[SpaceSort!])** `SpaceConnection`

The spaces available to this viewer.

**apiKeys** `[ApiKey!]`

The API keys belong to this viewer.

**annotationQueues(after: String, first: Int, before: String, last: Int, queueIds: \[ID!], search: String)** `ViewerAnnotationQueueConnection`

The queues assigned to the current viewer.

## Account

The account the caller belongs to. Account-level settings, users, audit logs, SAML configuration and LLM integrations hang off this type. The account the viewer belongs to

**id** `ID!`

The ID of an object

**createdAt** `DateTime!`

When the account was created.

**name** `String!`

The name of the account.

**organizations(after: String, first: Int, before: String, last: Int, search: String, sort: \[AccountOrganizationSort!])** `AccountOrganizationConnection`

The organizations under this account.

**accountOrganizationCount** `Int!`

The number of organizations under this account.

**tierLimits** `[AccountTierLimitsType!]`

The pricing tier's limits of the account

**signalIssueUsage** `SignalIssueUsage!`

The account's current-month Signal issue usage and per-tier cap. Drives the Signal upgrade CTA.

**monitorUsage** `MonitorUsage!`

**Deprecated:** Monitors are no longer capped per tier, so limit is always null and limitReached is always false. The account's monitor count across all spaces.

**users(after: String, first: Int, before: String, last: Int, search: String, sort: \[UserSort!], userType: UserType)** `UserConnection`

All users in this account.

**userSessionDurationMinutes** `Int!`

The number of minutes a user session will last before expiring.

**usage(startDate: DateTime, endDate: DateTime, datasourceType: DatasourceType, organizationId: ID, spaceName: String, externalModelId: String)** `[Usage!]`

The usage of the account

**stripeSession(stripeSessionId: String!)** `AccountStripeSessionType`

The Stripe session associated with the account based on a passed-in Stripe session ID. Will return null if no session found.

**planSubscriptionInfo** `AccountPlanSubscriptionInfo`

The current plan subscription associated with the account. Will return null if no subscription found.

**dataRetentionPeriod** `AccountRetention`

Data retention period for this account. Null means use plan default.

**dataRetentionCutoffDate** `DateTime`

The enforced UTC retention cutoff. Null means no account retention cutoff.

**effectiveDataRetentionPeriod** `AccountRetention`

Retention window enforced on queries after account exemptions. Null means no retention gate.

**authenticatedAuditLogs(after: String, first: Int, before: String, last: Int, startTime: DateTime, endTime: DateTime)** `AuthenticatedUserAuditLogConnection`

All authenticated mutations logged in this account.

**unauthenticatedAuditLogs(after: String, first: Int, before: String, last: Int, startTime: DateTime, endTime: DateTime)** `UnauthenticatedUserAuditLogConnection`

All unauthenticated mutations logged in this account - mostly login attempts

**exporterAuditLogs(after: String, first: Int, before: String, last: Int, startTime: DateTime, endTime: DateTime)** `ExporterAuditLogEntryConnection`

All exporter audit logs for this account.

**samlIdPs(after: String, first: Int, before: String, last: Int)** `SAMLIdPConnection`

SAML Identity Provider configurations for this account (active only)

**allSamlIdPs(after: String, first: Int, before: String, last: Int, includeDeleted: Boolean)** `SAMLIdPConnection`

All SAML Identity Provider configurations for this account (including soft-deleted)

**isSamlEnforced** `Boolean!`

Whether any SAML IdP on this account has SAML authentication enforcement enabled

**isSamlSyncEnabled** `Boolean!`

Whether any SAML IdP on this account has user role sync enabled

**developerAccessDefault** `Boolean!`

**Deprecated:** Developer access now comes from the assigned account role. Use developerAccessLocked to disable developer access for the account. Legacy developer access default. This field always returns true.

**developerAccessLocked** `Boolean!`

Whether developer access is disabled for all users in this account.

**llmIntegrations(includeAgentOnlyProviders: Boolean)** `[LlmIntegration!]`

LLM integrations configured for this account

**costConfigs(after: String, first: Int, before: String, last: Int, search: String, isDefault: Boolean)** `CostConfigConnection!`

Account-level cost configs, scope-filtered to what the viewer can read (RBAC scope-aware surface).

**internalLlmModels** `[InternalLlmModel!]`

Arize-managed internal models available for Alyx (Copilot).

**internalLlmProviders** `[InternalLlmProvider!]`

Arize-managed internal LLM providers available to managed agents (sandboxcontroller) in this deployment. Empty means no managed provider (on-prem / VPC) — callers must bring their own integration.

**sandboxIntegrations(provider: String)** `[SandboxIntegration!]`

Credentials a sandbox can use to authenticate to external services (e.g., GitHub).

**sandboxConfigs** `[SandboxConfig!]`

Reusable sandbox configurations for this account.

**slackAgentChannelBindings(after: String, first: Int, before: String, last: Int)** `SlackAgentChannelBindingConnection`

Slack channel bindings for this account: which channel a space's signal deliveries post to, and the models enabled for delivery. A Relay connection; realistic data is a handful of channels per account.

**slackAgentIntegrations** `[SlackAgentIntegration!]`

Connected Slack workspaces for this account's managed agents.

**sandboxProviders(provider: String)** `[SandboxProvider!]`

External sandbox provider configurations.

**claudeSandboxProvider(providerId: ID!)** `ClaudeSandboxProvider`

Claude managed metadata for a specific sandbox provider.

## AccountOrganization

An organization inside the account. Organizations hold spaces and organization-scoped integrations. An organization within an account that uses Arize.

**id** `ID!`

The ID of an object

**name** `String!`

The name of the account organization.

**createdAt** `DateTime!`

When the account organization was created.

**description** `String!`

A description of the account organization. May be empty.

**creator** `User`

The creator of the account organization.

**spaces(after: String, first: Int, before: String, last: Int, search: String, sort: \[SpaceSort!])** `SpaceConnection`

The spaces under this account organization.

**accountOrganizationUsers(after: String, first: Int, before: String, last: Int, search: String, userType: UserType)** `OrganizationUserConnection`

The users with access to the account organization.

**predictionVolume(startTime: DateTime!, endTime: DateTime!, environmentName: ModelEnvironmentName!)** `Float!`

The number of predictions across all models in an account organization in the given environment and time range.

**integrations(providerName: IntegrationProvider, search: String)** `[IntegrationKey!]!`

A list of all configured integrations for account organization

**externalLlmApiKeys(provider: ExternalLlmApiKeyProvider)** `[ExternalLlmApiKey!]`

API keys used to access external LLM's

**customLlmEndpoints** `[CustomLlmEndpoint!]`

Custom llm endpoints that the openai client can point to

**webhooks** `[Webhook!]`

Webhook destinations for this account organization (webhooks\_destinations).

**costConfigs(after: String, first: Int, before: String, last: Int, search: String, isDefault: Boolean)** `CostConfigConnection!`

The cost configs for an account organization

**alyxCustomViews(after: String, first: Int, before: String, last: Int, appliesTo: AlyxCustomViewAppliesTo!, annotationQueueId: ID)** `AlyxCustomViewConnection!`

Alyx-generated custom views that belong to this organization. Returns the union of the viewer's personal views and public views in the org.

**modelCount(ignoreDemoModels: Boolean!)** `[ModelTypeCount!]`

The number of models in an organization by model type

**monitorTriggeredCount(startDate: DateTime!, endDate: DateTime!)** `[MonitorCategoryCount!]`

The number of monitors triggered in an organization by monitor category over a period of time

**modelTriggeredCount(startDate: DateTime!, endDate: DateTime!)** `[ModelTriggeredCount!]`

The number of models triggered in an organization over a period of time

**modelInferencesCount(startDate: DateTime!, endDate: DateTime!)** `[ModelInferencesCount!]`

The number of model inferences for an account organization over a period of time

**llmSpansCount(startDate: DateTime!, endDate: DateTime!)** `[LlmSpansCount!]`

Count of LLM spans for an account organization over a period of time

**llmModelCosts(startDate: DateTime!, endDate: DateTime!)** `[LlmModelCosts!]`

LLM model costs for an account organization over a period of time

**llmModelCostsBySpace(startDate: DateTime!, endDate: DateTime!)** `[LlmModelCostsBySpace!]`

LLM model costs by space for an account organization over a period of time

**totalCostOverTime(startDate: DateTime!, endDate: DateTime!)** `[TotalCostOverTime!]!`

Total LLM cost per space per day for an account organization over a period of time

**averageLatencyOverTime(startDate: DateTime!, endDate: DateTime!)** `[AverageLatencyOverTime!]`

Average span latency per space per day for an account organization over a period of time

**evalNames(startDate: DateTime!, endDate: DateTime!)** `[EvalName!]!`

All eval names discovered across LLM models in the org

**averageEvalScoreOverTime(startDate: DateTime!, endDate: DateTime!, evalName: String, scope: EvalScope)** `[AverageEvalScoreOverTime!]!`

Average eval score per day across all models in the org for a specific eval

**evalLabelDistribution(startDate: DateTime!, endDate: DateTime!, evalName: String, scope: EvalScope)** `[EvalLabelCount!]!`

Distribution of eval labels across all models in the org for a specific eval

**traceCountOverTime(startDate: DateTime!, endDate: DateTime!)** `[CountOverTime!]!`

Trace count (root spans) per space per day for an account organization over a period of time

**errorsOverTime(startDate: DateTime!, endDate: DateTime!)** `[CountOverTime!]!`

Error span count (spans with status\_code=ERROR) per space per day for an account organization over a period of time

**summaryDashboardConfig** `SummaryDashboardConfig`

The summary dashboard config for an account organization

## Space

A workspace. Almost everything you manage through mutations (models and projects, monitors, dashboards, datasets, prompts, evaluators, tasks) is reached from a `Space`. An organization's workspace

**id** `ID!`

The ID of an object

**name** `String!`

The name of the space.

**description** `String!`

The description of the space.

**uuid** `String!`

The UUID of the space.

**createdAt** `DateTime!`

When the space was created.

**organization** `AccountOrganization!`

The organization that owns the space.

**liveEnabled** `Boolean!`

Whether live tracing is enabled for this space

**isLLMOnlySpace** `Boolean!`

Whether this space only contains LLM/tracing projects (no traditional ML models).

**playgroundTracingModelId** `ID`

The model ID for playground runs tracing in this space

**evalTracingModelId** `ID`

The model ID for evaluations tracing in this space

**sandboxTracingModelId** `ID`

The model ID for sandbox agent traces in this space

**mlModelsEnabled** `Boolean!`

Whether the ML Models tab is enabled for this space. When false, the tab is hidden regardless of model presence.

**private** `Boolean!`

**gradientStartColor** `String!`

Gradient start color

**gradientEndColor** `String!`

Gradient end color

**models(after: String, first: Int, before: String, last: Int, search: String, useExactSearchMatch: Boolean, tagsAny: \[ID!], tagsAll: \[ID!], sort: \[ModelSort!], filter: ModelConnectionFilterInputType)** `ModelConnection`

The models in this space

**monitors(after: String, first: Int, before: String, last: Int, search: String, monitorCategory: MonitorCategory, isManaged: Boolean, notificationsEnabled: Boolean, currentStatus: \[MonitorStatus!], usesCustomMetric: Boolean, modelName: String, modelType: ModelType, sort: \[MonitorSort!])** `MonitorConnection`

The monitors under this organization.

**dashboards(after: String, first: Int, before: String, last: Int, search: String, sort: \[DashboardSort!])** `DashboardConnection`

The dashboards under this organization.

**hasDashboardsCreated** `Boolean!`

Whether this space has any dashboards created.

**spaceUsers(after: String, first: Int, before: String, last: Int, search: String, userType: UserType)** `SpaceUserConnection!`

A full list of users in the account with access to this particular space. Includes space members, account admin, and account organization admin. Sorted by email. For only space members, see `spaceMembers`.

**byobConnectors(after: String, first: Int, before: String, last: Int)** `ByobConnectorConnection!`

The BYOB connectors in this space

**importJobs(after: String, first: Int, before: String, last: Int, sort: \[ImportJobSort!], search: String, cloudJobStatus: \[ImportJobStatus!], jobUploadState: JobUploadState, modelId: ID)** `FileImportJobConnection`

A listing of all the configured import jobs

**tableJobs(after: String, first: Int, before: String, last: Int, sort: \[ImportJobSort!], search: String, tableJobStatus: \[ImportJobStatus!], jobUploadState: JobUploadState, modelId: ID)** `TableImportJobConnection`

A listing for all the configured table import jobs

**sandboxJobs(after: String, first: Int, before: String, last: Int, automationId: ID, search: String, createdAfter: DateTime, excludeAutomationSpawned: Boolean, excludeHarnessEvaluation: Boolean)** `SandboxJobConnection!`

Managed agent sandbox jobs associated with this space

**signalInsightSummary** `SpaceSignalInsightSummary!`

Fleet-wide rollup of Signal issue activity across every project in this space. Powers the Swarm Manager page's Fleet summary widget.

**signalEnabledProjectSummary** `SignalEnabledProjectSummary!`

Exact count of projects with Signal enabled, out of every project the caller can read in this space. Powers the Swarm Manager page's Fleet summary widget.

**signalIssues(limit: Int)** `[SignalIssue!]!`

Open (and ignored) Signal issues across every project in this space, no date filter — the current backlog, matching signalInsightSummary.openCount's semantics. Ignored issues sort last; active issues are ordered by severity then recency. Powers the Swarm Manager page's 'Insights from Signal' section.

**agentIssues(limit: Int)** `[SignalIssue!]!`

Open (and ignored) issues across every issue-producing agent (Signal, Cost, ...) in this space — the producer-inclusive umbrella backing the 'Issues' view. Unlike signalIssues (Signal-only), this spans all producers; each issue carries its `producer` for a per-agent badge. Same backlog/ordering semantics as signalIssues.

**automations** `[Automation!]!`

Managed agent automations (cron- or metric-triggered sandbox job kickoffs) configured for this space.

**datasets(after: String, first: Int, before: String, last: Int, search: String, tagsAny: \[ID!], tagsAll: \[ID!], sort: \[DatasetSort!])** `DatasetConnection`

The datasets in this space

**experiments(after: String, first: Int, before: String, last: Int, search: String, tagsAny: \[ID!], tagsAll: \[ID!], sort: \[ExperimentSort!], groupBy: ExperimentGroupBy, sources: \[ExperimentSourceType!])** `SpaceExperimentConnection`

All active experiments in this space, grouped and ordered server-side.

**customMetrics(after: String, first: Int, before: String, last: Int, searchTerm: String, sort: \[ModelCustomMetricSort!])** `CustomMetricConnection`

All custom metrics across all models in this space.

**onlineTasks(after: String, first: Int, before: String, last: Int, sort: \[OnlineTaskSort!], filter: OnlineTaskConnectionFilterInputType, modelId: ID, datasetId: ID, experimentIds: \[ID!], search: String)** `OnlineTaskConnection`

The online tasks in this space

**prompts(after: String, first: Int, before: String, last: Int, search: String, name: String, versionId: ID, versionLabel: String, tagsAny: \[ID!], sort: \[PromptSort!])** `PromptConnection`

The prompts in this space

**tags(after: String, first: Int, before: String, last: Int, search: String, sort: \[TagSort!])** `TagConnection`

The tags in this space

**evaluators(after: String, first: Int, before: String, last: Int, search: String, name: String, sort: \[EvaluatorSort!], taskType: OnlineTaskType, tagsAny: \[ID!], hasOnlineTasks: Boolean)** `EvaluatorConnection`

The evaluators in this space

**annotationQueues(after: String, first: Int, before: String, last: Int, search: String, userId: ID)** `AnnotationQueueConnection`

The annotation queues in this space

**playgroundViews(after: String, first: Int, before: String, last: Int, sort: \[PlaygroundViewSort!], sources: \[PlaygroundViewSource!])** `PlaygroundViewConnection`

The playground views in this space

**annotationConfigs(after: String, first: Int, before: String, last: Int, search: String)** `SpaceAnnotationConfigConnectionConnection`

The annotation configs in this space

**traceFilters(after: String, first: Int, before: String, last: Int, search: String, modelId: ID!, modelEnvironmentName: ModelEnvironmentName!)** `SpaceTraceFilterConnectionConnection`

The trace filters in this space

**llmIntegrations(includeAgentOnlyProviders: Boolean)** `[LlmIntegration!]`

LLM integrations available to this space (includes space-scoped, org-wide, and account-wide integrations)

**defaultLlmIntegration** `SpaceDefaultLlmIntegration`

The default LLM integration and model configured for this space, or null if none is set.

**defaultAgentRuntime** `LlmIntegration`

The default agent runtime (an agent-eligible Anthropic AI Provider) for this space, used to seed the runtime picker for new agents. Null means the Arize-managed model.

**gates(gateType: TierGate)** `[AccountTierLimitsType!]`

Rate limit gates for this space.

**\_access** `ViewerAccess!`

## Model

A model or LLM project. `projectType` tells you which. Monitors, dashboards, metrics, tracing stats and Signal issues are reached from here. A model on the Arize platform

**id** `ID!`

The ID of an object

**name** `String!`

The name of the model.

**uri** `URL!`

The URI of the model in the Arize platform.

**createdAt** `DateTime!`

The creation date of the model.

**isDemoModel** `Boolean!`

Whether the model is a demo model.

**isRestricted** `Boolean!`

Whether this project has restricted access. When true, only users with explicit project-level role bindings (or space/account admins) can access it.

**projectRoleBindings** `[RoleBinding!]!`

The project-level role bindings for this model. Empty when the project is not restricted.

**space** `Space!`

The space that the model belongs to

**volumeGateVisibility** `VolumeGateVisibility!`

Whether tracing content for this project should be hidden due to a volume/storage gate being exceeded. The single source of truth shared by Signal, Evals & Metrics, Embeddings, Agent Path/Graph, and Traces/Spans/Sessions row-hiding.

**modelVersions(after: String, first: Int, before: String, last: Int, searchTerm: String, environmentName: ModelEnvironmentName)** `ModelVersionConnection`

All the versions of this model.

**modelBatches(after: String, first: Int, before: String, last: Int, sort: \[ModelDatasetSort!])** `ModelBatchConnection`

The batches associated with all the versions and environments on this model.

**modelPredictionVolume(startTime: DateTime, endTime: DateTime, timeZone: String)** `ModelPredictionVolume`

The number of predictions seen for this model in the time range. Defaulted to last 30 days.

**modelType** `ModelType!`

The type of the model.

**modelEnvironments** `[ModelEnvironmentListObj]`

All the environments of this model. Including the bookends of each environment

**modelEnvironmentNames** `[ModelEnvironmentName!]!`

All the names of the environments of this model (i.e., environments where the model has data).

**dimension(startTime: DateTime!, endTime: DateTime!, name: String!, category: DimensionCategory!, environmentName: ModelEnvironmentName)** `DimensionDetails`

The details of a dimension (feature, prediction, or actual). A way to navigate directly to a dimension without paginating through a list.

**dimensionValues(after: String, first: Int, before: String, last: Int, dimension: DimensionInput, modelEnvironmentName: ModelEnvironmentName, filterType: FilterRowType, searchTerm: String)** `DimensionValueConnection`

The values of a dimension.

**tracingSchema(startTime: DateTime, endTime: DateTime)** `TracingSchemaType`

The set of properties shared by the LLM spans in a given time range.

**modelSchema(startTime: DateTime, endTime: DateTime, modelEnvironmentName: ModelEnvironmentName, modelVersionEnvironmentMetadataIds: \[ID!], externalModelVersionIds: \[String!])** `ModelSchema`

The schema of a model.

**modelPrimaryBaseline** `ModelBaseline`

The model's primary baseline.

**modelReferenceDataset(timestamp: DateTime)** `ModelReferenceDataset`

The reference dataset to be used for analysis - e.g. drift and performance.

**modelDashboards(after: String, first: Int, before: String, last: Int, search: String, sort: \[ModelDashboardSort!])** `ModelDashboardConnection`

A list of dashboards that reference this model.

**monitors(after: String, first: Int, before: String, last: Int, search: String, creationStatus: MonitorCreationStatus, monitorCategory: MonitorCategory, isManaged: Boolean, notificationsEnabled: Boolean, usesCustomMetric: Boolean, customMetricId: ID, currentStatus: \[MonitorStatus!], sort: \[MonitorSort!], dimensionCategory: DimensionCategory, dataQualityMetric: DataQualityMetric)** `MonitorConnection`

The monitors on this model.

**signalStatus** `SignalStatus!`

The Signal automation status for this project — state, schedule, and the preset it runs.

**signalIssues(startDate: DateTime, endDate: DateTime, limit: Int, producer: String)** `[SignalIssue!]!`

Issues surfaced by the project's Signal automation (or another issue-filing agent, via `producer`), deduplicated across every cron-fired run that reported them under the same `issueKey`. Scoped to issues active (last seen) in the window when one is given, so recurred issues surface. Excludes user-resolved issues; open issues come first, then ignored ones (sorted to the bottom for the muted group). Returns \[] when the given producer's automation is not configured for this project.

**signalIssueActivityBounds** `SignalIssueActivityBounds`

The activity bounds across the project's live Signal issues (`earliest` = oldest first-seen, `latest` = most recent last-seen), or null when none exist. Lets the FE default the Signal tab's time window to recent Signal activity, covering issues that are still actively recurring. Returns null when Signal is not configured for this project.

**signalIssueCountsByEntity(startDate: DateTime!, endDate: DateTime!)** `[SignalIssueEntityCount!]!`

Distinct issue counts grouped by `entity` for the project's Signal automation. Powers the 'Issues found' donut.

**signalIssueTimeSeries(startDate: DateTime!, endDate: DateTime!, bucket: SignalIssueBucketSize!, timeZone: String)** `[SignalIssueTimeBucket!]!`

Time-bucketed counts of new issues (by first-seen time) grouped by `entity` for the project's Signal automation. Powers the 'Issues over time' stacked bar.

**monitorAggregations** `MonitorAggregationsType`

The monitor metrics of this model, ex. number of monitors triggered.

**driftConfigs** `ModelDriftConfigs!`

The drift configs for this model, such as default drift metric and evaluation window.

**performanceConfigs** `ModelPerformanceConfigs!`

The performance configs for this model, such as default performance metric and prediction class.

**dataQualityConfigs** `ModelDataQualityConfigs!`

The data quality configs for this model.

**modelColumnMappingConfigs** `ModelColumnMappingConfig`

The column mapping configs for this model such as output --> attributes.llm.output\_messages.0.content

**projectSourceMappingConfigs** `ProjectSourceMappingConfig`

Source mapping configs that map target columns (input/output) to JSON attribute paths for ingestion-time extraction.

**performanceMetrics(startDate: DateTime!, endDate: DateTime!)** `ModelPerformanceMetrics!`

The performance metrics for the models in the project.

**performanceStatus** `ModelPerformanceStatus!`

The aggregate performance status of a model.

**averagePerformanceMetric(dataset: ModelDatasetInput, modelBaselineId: ID, baselineFilters: \[FilterItemInputType!], movingBaselineStartTime: DateTime, movingBaselineEndTime: DateTime, performanceMetric: PerformanceMetric!, atRankingKValue: PositiveInt, positiveClass: String, customMetricConfig: String, timeZone: String!)** `Float`

The average performance metric value across of the model baseline OR dataset.

**performanceMetricOverTime(startTime: DateTime!, endTime: DateTime!, timeZone: String!, environmentName: ModelEnvironmentName!, externalModelVersionIds: \[String!], externalBatchIds: \[String!], performanceMetric: PerformanceMetric!, atRankingKValue: PositiveInt, positiveClass: String, filters: \[FilterItemInputType!], queryFilter: String, fairnessMetricConfig: FairnessMetricConfigInput, customMetricConfig: String, evaluationWindowMs: Float, timeSeriesDataGranularity: DataGranularity!)** `ModelPerformanceMetricOverTimeType!`

The metric of the model's prediction over time for a specific evaluation metric, ex. accuracy, prediction, etc.

**customMetrics(after: String, first: Int, before: String, last: Int, searchTerm: String, sort: \[ModelCustomMetricSort!])** `CustomMetricConnection`

All the custom metric functions on the model.

**driftStatus** `ModelDriftStatus!`

The aggregate drift status of the model.

**dataQualityStatus** `ModelDataQualityStatus!`

The aggregate data quality status of the model.

**totalTraffic(dataset: ModelDatasetInput, modelBaselineId: ID, baselineFilters: \[FilterItemInputType!], movingBaselineStartTime: DateTime, movingBaselineEndTime: DateTime, timeZone: String!, tracesOnly: Boolean, sessionsOnly: Boolean)** `Int!`

The total traffic of a given model over a given period of time. A value of 0 means there is no data.

**UMAPPoints(dataset: ModelDatasetInput!, referenceDataset: ModelDatasetInput, modelBaselineId: ID, baselineFilters: \[FilterItemInputType!], nComponents: Int!, nNeighbors: Int!, minDist: Float!, minClusterSize: Int!, clusterSelectionEpsilonRatio: Float!, datasetSampleSize: Int!, referenceDatasetSampleSize: Int, embeddingDimensions: \[DimensionInput!]!)** `UmapPointsOrError!`

The UMAP projection of the dimension for both a primary and reference dataset

**spanRecordsPublic(after: String, first: Int, before: String, last: Int, dataset: ModelDatasetInput!, columnNames: \[String!], sort: SpanSort, includeRootSpans: Boolean, truncateStringLength: Int, includeColumnMapping: Boolean)** `SpanRecordConnection`

A list containing span ID's and user-specified columns with their respective row values

**costOverTime(startDate: DateTime!, endDate: DateTime!)** `[ModelCostOverTime!]!`

This project's total LLM inference cost per day over a given window. Powers the Cost tab's cost-over-time chart.

**tokensOverTime(startDate: DateTime!, endDate: DateTime!)** `[ModelTokensOverTime!]!`

This project's prompt vs. completion token usage per day over a given window. Powers the Cost tab's tokens-over-time chart.

**costByModel(startDate: DateTime!, endDate: DateTime!)** `[ModelCostByLlmModel!]!`

This project's total LLM inference cost broken down by model over a given window. Powers the Cost tab's cost-by-model chart.

**llmTracingStats(dataset: ModelDatasetInput!, timeZone: String!)** `LlmTracingStats`

The LLM tracing stats of the model, must be used with the tracing environment.

**taskVariables(queryFilter: String, includeColumnMapping: Boolean)** `[String!]!`

The task variables for the model

**annotationsEnabled** `Boolean!`

Whether annotations are enabled for the model.

**isAgentExperimentTraceProject** `Boolean!`

Whether this project is an agent experiment trace project. These are hidden from the Tracing Projects list.

**projectType** `ProjectType`

The project type for generative LLM projects. Null for classic ML models.

**annotationsDisabled** `Boolean!`

Whether annotations are disabled for the model.

**network(dataset: ModelDatasetInput!, nodeIdDimensionName: String!, parentNodeDimensionName: String, nodeStepDimensionName: String, nodeNameDimensionName: String)** `ModelNetwork!`

Network edges of spans for agent visualization

**tags(after: String, first: Int, before: String, last: Int)** `TagConnection`

The tags associated with the model

**defaultAnnotationConfigIds** `[ID!]!`

The default annotation config ids for the model

**\_access** `ViewerAccess!`