toto-models 1.0.0

Umbrella package for Datadog Toto time series models

toto-models

GitHub | Toto 2.0 Report | Toto 2.0 Blog | Model Collection | BOOM Dataset

toto-models is the recommended way to install Datadog's Toto time series models. It is an umbrella package that pulls in toto-2 and its dependencies so you don't have to manage them individually.

Installation

pip install toto-models

To also install Toto 1.0 (for fine-tuning or exogenous variable support, not yet available in 2.0):

pip install "toto-models[v1]"

Quick Start

import torch
from toto2 import Toto2Model

model = Toto2Model.from_pretrained("Datadog/Toto-2.0-22m")
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model = model.to(device).eval()

# Input shape: (batch, n_variates, time_steps)
target = torch.randn(1, 1, 512, device=device)
target_mask = torch.ones_like(target, dtype=torch.bool)
series_ids = torch.zeros(1, 1, dtype=torch.long, device=device)

# Returns quantiles of shape (9, batch, n_variates, horizon)
quantiles = model.forecast(
    {"target": target, "target_mask": target_mask, "series_ids": series_ids},
    horizon=96,
    decode_block_size=768,
    has_missing_values=False,
)

For full documentation, see the toto-2 package or the GitHub repository.

Citation

@misc{khwaja2026toto20timeseries,
      title={Toto 2.0: Time Series Forecasting Enters the Scaling Era},
      author={Emaad Khwaja and Chris Lettieri and Gerald Woo and Eden Belouadah and Marc Cenac and Guillaume Jarry and Enguerrand Paquin and Xunyi Zhao and Viktoriya Zhukov and Othmane Abou-Amal and Chenghao Liu and Ameet Talwalkar and David Asker},
      year={2026},
      eprint={2605.20119},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2605.20119},
}

License

Apache-2.0. See LICENSE for details.