Contamination MapTime series / Model
Toto 2.0
- Organisation
- Datadog
- Family
- Toto · version 2.0
- Released
- 2026-05-19 (arXiv v1 submission date of the technical report (https://arxiv.org/abs/2605.20119); HF repos created 2026-04-14 (4m/22m/313m/1B) and 2026-04-17 (2.5B) per HF API createdAt.)
- Checkpoints
- Datadog/Toto-2.0-4m, Datadog/Toto-2.0-22m, Datadog/Toto-2.0-313m, Datadog/Toto-2.0-1B, Datadog/Toto-2.0-2.5B
- Weights
- Open · apache-2.0
- Links
- PaperModel cardCode
Training data disclosure: Vague
Level determined for 75% of benchmark cells.
- TestTest data seen0
- SeenSeen, split unclear0
- TrainTraining data seen0
- Related data seen0
- NoNot seen37
- ?Unknown12
“Toto 2.0 trains exclusively on a mix of Datadog's internal telemetry and synthetic data.”
“The final training mix for base models is 42.5% observability data and 57.5% synthetic data”
“Toto 2.0 uses the synthetic data generation method from TempoPFN (Moroshan et al., 2025)”
Declared exclusions
-
Benchmark GIFT-Eval · entirely excluded
Evidence
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
arxiv.org · paper · Section 1 · retrieved 2026-10-15 “First, we removed all public data from pretraining.”
arxiv.org · paper · Section 3 · retrieved 2026-10-15 “The base checkpoints are pretrained without any public data”
huggingface.co · model card · model card warning · retrieved 2026-10-15 -
Benchmark Long-horizon forecasting (ETT, Electricity, Traffic, Weather, Exchange, ILI) · entirely excluded
Evidence
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
arxiv.org · paper · Section 1 · retrieved 2026-10-15
Benchmark cells
GIFT-Eval
Not seen Jena Weather (GIFT-Eval) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen BizITObs - Application · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen BizITObs - Service · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen BizITObs - L2C · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Bitbrains - Fast Storage · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Bitbrains - rnd · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Restaurant (Recruit Restaurant Visitor Forecasting) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen ETT1 (GIFT-Eval) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen ETT2 (GIFT-Eval) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Loop Seattle (LibCity) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen SZ-Taxi (LibCity) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen M_DENSE (LibCity) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Solar (LSTNet, 137 Alabama PV plants) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Hierarchical Sales (Mancuso et al., 2020) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen M4 Yearly · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen M4 Quarterly · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen M4 Monthly · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen M4 Weekly · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen M4 Daily · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen M4 Hourly · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Hospital · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen COVID Deaths · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen US Births · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Saugeen River Flow · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Temperature Rain · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen KDD Cup 2018 · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen Car Parts · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Not seen ElectricityLoadDiagrams20112014 (UCI) · the authors state the dataset was excluded from training
Evidence (3)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
“First, we removed all public data from pretraining.”
“The base checkpoints are pretrained without any public data”
Long-horizon forecasting (ETT, Electricity, Traffic, Weather, Exchange, ILI)
Not seen ETTh1 · the authors state the dataset was excluded from training
Evidence (1)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
Not seen ETTh2 · the authors state the dataset was excluded from training
Evidence (1)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
Not seen ETTm1 · the authors state the dataset was excluded from training
Evidence (1)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
Not seen ETTm2 · the authors state the dataset was excluded from training
Evidence (1)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
Not seen Electricity (ECL, 321 clients) · the authors state the dataset was excluded from training
Evidence (1)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
Not seen Traffic (PeMS San Francisco Bay, 862 sensors) · the authors state the dataset was excluded from training
Evidence (1)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
Not seen Weather (Jena, 2020; Autoformer) · the authors state the dataset was excluded from training
Evidence (1)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
Not seen Exchange-Rate · the authors state the dataset was excluded from training
Evidence (1)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
Not seen ILI (CDC influenza-like illness, national) · the authors state the dataset was excluded from training
Evidence (1)
“Unlike other leading TSFMs, we do not pretrain on any public time series data, and instead rely exclusively on a mix of Datadog's internal observability metrics and synthetic data.”
Time-series anomaly detection
Unknown TSB-AD-U (TSB-AD univariate) · training data not disclosed in enough detail
Unknown TSB-AD-M (TSB-AD multivariate) · training data not disclosed in enough detail
Unknown UCR Time Series Anomaly Archive (Hexagon ML/UCR, KDD Cup 2021) · training data not disclosed in enough detail
Unknown Server Machine Dataset (SMD) · training data not disclosed in enough detail
Unknown Mars Science Laboratory (MSL) telemetry anomaly dataset · training data not disclosed in enough detail
Unknown Soil Moisture Active Passive (SMAP) telemetry anomaly dataset · training data not disclosed in enough detail
Unknown Pooled Server Metrics (PSM) · training data not disclosed in enough detail
Unknown Secure Water Treatment (SWaT) · training data not disclosed in enough detail
Unknown Yahoo S5 Labeled Anomaly Detection Dataset (Yahoo Webscope S5) · training data not disclosed in enough detail
Unknown Numenta Anomaly Benchmark (NAB) · training data not disclosed in enough detail
Time-series classification
Unknown UCR Time Series Classification Archive (2018) · training data not disclosed in enough detail
Unknown UEA Multivariate Time Series Classification Archive (2018) · training data not disclosed in enough detail
Notes
Five sizes share one recipe: 42.5% Datadog internal observability metrics (10s/60s/5m+ intervals; no customer data) + 57.5% TempoPFN synthetic; 5.04T points (313m/1B/2.5B) or 3.40T (4m/22m). No public data in pretraining, so no overlap with GIFT-Eval or long-horizon datasets is possible from the stated recipe. Disclosure marked vague because the real-data part is an undisclosed internal source (no named datasets). No BOOM leakage statement in the Toto 2.0 report.
Sources
- Toto 2.0: Time Series Forecasting Enters the Scaling Era (arXiv v2) · paper · retrieved 2026-10-15
- Toto-2.0-2.5B model card · model card · retrieved 2026-10-15
- Toto-2.0-4m model card · model card · retrieved 2026-10-15
- Toto-2.0-2.5B-FT model card · model card · retrieved 2026-10-15