toto 2.2.0.0

DataDog Toto 2.0 — a foundation model for multivariate time series forecasting with a focus on observability metrics. Use when working with zero-shot time series forecasting, probabilistic predictions (quantile-based), multi-variate time series, or observability metric prediction. Covers the full Toto 2.0 model family (4m to 2.5B parameters), the Toto2Model API, forecast(), decode strategies (single-pass and block decode), GluonTS integration via Toto2GluonTSModel, BOOM and GIFT-Eval benchmarks, and missing-value handling. Default model is Datadog/Toto-2.0-22m. Triggers on toto, toto-2, time-series forecasting, observability metrics, zero-shot forecasting, quantile foreca…