Forecasting APIby Birla AI Labs
Get API key

Data Schemas

Complete type definitions for all request and response objects.

InputSeries

One time series submitted for forecasting.

targetarray[number]requiredFinite values to forecast.
indexarray[string]requiredStrictly ascending ISO dates or datetimes matching target length.
hist_variablesobject<string, array[number]>optionalHistorical covariates; every array matches target length.
future_variablesobject<string, array[number]>optionalFuture covariates; every array matches horizon.
future_variables_indexarray[string]optionalFuture-covariate timestamps matching horizon.
static_variablesobjectoptionalScalar covariates for the series.

ForecastRequest

Body shared by synchronous and asynchronous forecasting.

modelstringrequiredModel ID or supported alias. Toto checkpoints use toto-2 or toto-2-ft.
targetarray[number]requiredFinite values to forecast.
indexarray[string]requiredStrictly ascending ISO dates or datetimes matching target length.
hist_variablesobject<string, array[number]>optionalHistorical covariates; every array matches target length.
future_variablesobject<string, array[number]>optionalFuture covariates; every array matches horizon.
future_variables_indexarray[string]optionalFuture-covariate timestamps matching horizon.
static_variablesobjectoptionalScalar covariates for the series.
horizonintegerrequiredForecast steps, bounded by the selected model.
freqfrequencyrequired5min, 15min, 30min, H, D, W, M, MS, Q, or Y.
contextintegeroptionalHistory steps supplied to the model.
quantilesarray[number]optionalValues strictly between 0 and 1. Defaults to every decile from 0.1 through 0.9.
model_sizestringoptionalSize for a multi-size model. Both Toto-2 checkpoints use the fixed 2.5b size.
modelstringModel implementation selected for the override.
configobjectConfiguration values passed to the selected model.
seedintegeroptionalReproducibility seed; default 42.
allow_truncationbooleanoptionalAllow context-window trimming; default false.

ForecastResponse

A completed forecast returned immediately or through job polling.

statuscompletedrequired
modelstringrequired
predictionobject<string, array[number]>required
indexarray[string]required
warningsarray[ForecastWarning] | nullStable warnings such as context truncation.

JobSubmitResponse

An accepted forecast queued for asynchronous execution.

job_idstringrequired
statusqueuedrequired
modelstringrequired
poll_urlstringrequiredSame-API URL used to poll the job.

JobStatusResponse

The current state or result of an asynchronous job.

statusqueued | in_progress | completed | failed | cancelledrequired
modelstring | null
seriesarray[array[SeriesPrediction]] | null
result_urlstring | nullSigned URL for a large completed result.
result_expires_atstring | null
errorstring | null

ModelInfo

A model returned by the public model registry.

model_idstringrequired
namestringrequired
descriptionstringrequired
sizesarray[string]required
supports_covariatesbooleanrequired
Example
{
  "index": [
    "2024-01-01",
    "2024-02-01",
    "2024-03-01"
  ],
  "target": [120.0, 126.0, 133.0],
  "hist_variables": {
    "temperature": [20.0, 21.0, 22.0]
  },
  "future_variables": {
    "temperature": [20.0, 21.0, 22.0, 23.0]
  },
  "future_variables_index": [
    "2024-01-01",
    "2024-02-01",
    "2024-03-01",
    "2024-04-01"
  ],
  "static_variables": {
    "region_id": 42
  }
}

Data Schemas

Complete type definitions for all request and response objects.

InputSeries

One time series submitted for forecasting.

targetarray[number]requiredFinite values to forecast.
indexarray[string]requiredStrictly ascending ISO dates or datetimes matching target length.
hist_variablesobject<string, array[number]>optionalHistorical covariates; every array matches target length.
future_variablesobject<string, array[number]>optionalFuture covariates; every array matches horizon.
future_variables_indexarray[string]optionalFuture-covariate timestamps matching horizon.
static_variablesobjectoptionalScalar covariates for the series.

ForecastRequest

Body shared by synchronous and asynchronous forecasting.

modelstringrequiredModel ID or supported alias. Toto checkpoints use toto-2 or toto-2-ft.
targetarray[number]requiredFinite values to forecast.
indexarray[string]requiredStrictly ascending ISO dates or datetimes matching target length.
hist_variablesobject<string, array[number]>optionalHistorical covariates; every array matches target length.
future_variablesobject<string, array[number]>optionalFuture covariates; every array matches horizon.
future_variables_indexarray[string]optionalFuture-covariate timestamps matching horizon.
static_variablesobjectoptionalScalar covariates for the series.
horizonintegerrequiredForecast steps, bounded by the selected model.
freqfrequencyrequired5min, 15min, 30min, H, D, W, M, MS, Q, or Y.
contextintegeroptionalHistory steps supplied to the model.
quantilesarray[number]optionalValues strictly between 0 and 1. Defaults to every decile from 0.1 through 0.9.
model_sizestringoptionalSize for a multi-size model. Both Toto-2 checkpoints use the fixed 2.5b size.
modelstringModel implementation selected for the override.
configobjectConfiguration values passed to the selected model.
seedintegeroptionalReproducibility seed; default 42.
allow_truncationbooleanoptionalAllow context-window trimming; default false.

ForecastResponse

A completed forecast returned immediately or through job polling.

statuscompletedrequired
modelstringrequired
predictionobject<string, array[number]>required
indexarray[string]required
warningsarray[ForecastWarning] | nullStable warnings such as context truncation.

JobSubmitResponse

An accepted forecast queued for asynchronous execution.

job_idstringrequired
statusqueuedrequired
modelstringrequired
poll_urlstringrequiredSame-API URL used to poll the job.

JobStatusResponse

The current state or result of an asynchronous job.

statusqueued | in_progress | completed | failed | cancelledrequired
modelstring | null
seriesarray[array[SeriesPrediction]] | null
result_urlstring | nullSigned URL for a large completed result.
result_expires_atstring | null
errorstring | null

ModelInfo

A model returned by the public model registry.

model_idstringrequired
namestringrequired
descriptionstringrequired
sizesarray[string]required
supports_covariatesbooleanrequired