Toto 1.0 is a legacy hosted model. TSFM.ai plans to deprecate it for new zero-shot usage after Toto 2.0 is validated in production; keep using 1.0 only if you depend on its upstream fine-tuning or exogenous-variable workflow. View the recommended replacement.
Datadog logo

Toto-Open-Base-1.0

Legacyonline
Datadog/Toto-Open-Base-1.0

151M params | 4K context | $0.00025 per forecast

Toto-Open-Base-1.0 is Datadog's first open observability-oriented forecasting foundation model, a 151M-parameter checkpoint purpose-built for the sparse, high-dimensional telemetry that infrastructure monitoring produces. It is the original Toto 1.0 release and a legacy hosted model on TSFM.ai: still useful for workflows that depend on its upstream fine-tuning or exogenous-variable path, but no longer the default for new zero-shot work.

Architecturally it is a decoder-only transformer with proportional factorized space-time attention and a Student-T mixture output head, the latter giving it native probabilistic, multivariate forecasts over correlated metric streams. Pretraining spans over 2T points total: roughly 1T internal observability metrics, public GiftEvalPretrain and Chronos data, plus synthetic series, and the official card states that no customer data was used.

On TSFM.ai, keep using Toto 1.0 only if you rely on its upstream fine-tuning or exogenous-variable workflow; the platform plans to deprecate it for new zero-shot usage once Toto 2.0 is validated in production. For new observability forecasting, start with the recommended Datadog/Toto-2.0-313m, which carries the newer u-muP-scaled architecture and quantile interface and is the designated replacement for this checkpoint.

Model Classification

Family

Toto

Type

time series foundation model

Pretrained time-series model exposed on TSFM.ai for zero-shot or few-shot forecasting workloads.

Training Data

Over 2T points total: roughly 1T internal observability metrics, public GiftEvalPretrain and Chronos data, plus synthetic series; the official card states no customer data was used.

Recommended For

  • Infrastructure, observability, and telemetry forecasting
  • Sparse, noisy, high-dimensional operational metrics

Strengths

  • Built around real observability-like workloads rather than only clean academic datasets
  • Strong benchmark fit for BOOM-style evaluation

Limitations

  • More specialized than general-purpose forecasting families
  • May be less intuitive as a default pick for simple low-dimensional business series
  • Legacy generation: use Toto 2.0 for new zero-shot observability workloads unless you need Toto 1.0 fine-tuning or exogenous-variable support

Capabilities

forecastingprobabilistic-forecastingmultivariateobservability

Tags

datadogobservabilitymultivariateprobabilisticlegacy

Specifications

Parameters
151M
Architecture
decoder-only transformer with proportional factorized space-time attention and Student-T mixture output
Context length
4,096
Max context
4,096
Minimum history
n/a
Recommended history
n/a
Input step
n/a
Required target series
1
Temperature
Ignored
Top P
Ignored
Max output
1,024
Avg latency
n/a
Uptime
n/a
Plan limits
1,000 rpm free · 1,000,000 rpm with billing
Accelerator
T4
Regions
Virginia, US
License
n/a

Pricing

Per forecast
$0.00025

Performance

Average latency
n/a
Availability
n/a
Plan limits
1,000 rpm free · 1,000,000 rpm with billing

Related models