Chronos-2: A foundation model for multivariate time series forecasting
Abstract
Accurate forecasting with uncertainty quantification is critical in a variety of scientific applications ranging from epidemiology and economics to energy systems and earth sciences. Yet most forecasting models still need to be developed and trained separately for each domain-specific dataset. Time series foundation models (TSFMs) aim to remove this limitation and enable the application of the same model across domains through large-scale pretraining. However, most existing TSFMs treat each time series independently and therefore cannot model interactions among time series and external drivers. Here, we introduce Chronos-2, a foundation model for zero-shot forecasting that models the interactions between multivariate time series and heterogeneous covariates, which commonly arise in scientific applications. The model infers relationships among time series directly from historical context and generates probabilistic forecasts without task-specific treatment. Across various benchmarks, Chronos-2 achieves state-of-the-art accuracy on average by improving skill scores by 5.2 percentage points relative to the strongest baselines and increasing win rates by 10 percentage points. Chronos-2 is also computationally efficient, generating forecasts for 100 time series in under one second on a single mid-range GPU, 50x and 1000x faster end-to-end than the best task-specific point and probabilistic forecasters, respectively. In addition, we demonstrate its effectiveness in diverse scientific settings -- including infectious-disease forecasting, macroeconomic modeling, carbon-flux prediction, and runoff simulation -- where the model matches or surpasses specialized methods developed for each domain. These results show that a single pretrained model can capture structure shared across temporal systems.
Related
- Improved Forecasting for Multivariate Time Series using GNAR models with General Inverse Distance Weights
- GNA: Granular Neighbor Assembly for Retrieval-Augmented Multivariate Time-Series Forecasting
- SGA: Uncertainty Quantification for Multi-Step Forecasting in Time Series Foundation Models
- RACE: Residual-Aware Test-Time Adaptation for Neighbor-Rich Time-Series Foundation Model Forecasting
- Evaluating Time-Series Foundation Models and Multimodal Dietary Context for CGM Forecasting
BibTeX
@misc{ansari2026chronos2,
title = {{Chronos-2: A foundation model for multivariate time series forecasting}},
author = {Abdul Fatir Ansari and Oleksandr Shchur and Jaris Küken and Andreas Auer and Boran Han and Pedro Mercado and Syama Sundar Rangapuram and Huibin Shen and Lorenzo Stella and Xiyuan Zhang and Mononito Goswami and Shubham Kapoor and Danielle C. Maddix and Pablo Guerron and Florian Saupe and Markus Reichstein and Tony Hu and Junming Yin and Nick Erickson and Prateek Mutalik Desai and Hao Wang and Huzefa Rangwala and George Karypis and Yuyang Wang and Michael Bohlke-Schneider},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.41956},
eprint = {2610.41956},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.41956},
}
RIS
TY - UNPB
TI - Chronos-2: A foundation model for multivariate time series forecasting
AU - Abdul Fatir Ansari
AU - Oleksandr Shchur
AU - Jaris Küken
AU - Andreas Auer
AU - Boran Han
AU - Pedro Mercado
AU - Syama Sundar Rangapuram
AU - Huibin Shen
AU - Lorenzo Stella
AU - Xiyuan Zhang
AU - Mononito Goswami
AU - Shubham Kapoor
AU - Danielle C. Maddix
AU - Pablo Guerron
AU - Florian Saupe
AU - Markus Reichstein
AU - Tony Hu
AU - Junming Yin
AU - Nick Erickson
AU - Prateek Mutalik Desai
AU - Hao Wang
AU - Huzefa Rangwala
AU - George Karypis
AU - Yuyang Wang
AU - Michael Bohlke-Schneider
PY - 2026
DA - 2026/10/30
PB - arXiv
DO - 10.48550/arxiv.2610.41956
UR - https://arxiv.org/abs/2610.41956
AB - Accurate forecasting with uncertainty quantification is critical in a variety of scientific applications ranging from epidemiology and economics to energy systems and earth sciences. Yet most forecasting models still need to be developed and trained separately for each domain-specific dataset. Time series foundation models (TSFMs) aim to remove this limitation and enable the application of the same model across domains through large-scale pretraining. However, most existing TSFMs treat each time series independently and therefore cannot model interactions among time series and external drivers. Here, we introduce Chronos-2, a foundation model for zero-shot forecasting that models the interactions between multivariate time series and heterogeneous covariates, which commonly arise in scientific applications. The model infers relationships among time series directly from historical context and generates probabilistic forecasts without task-specific treatment. Across various benchmarks, Chronos-2 achieves state-of-the-art accuracy on average by improving skill scores by 5.2 percentage points relative to the strongest baselines and increasing win rates by 10 percentage points. Chronos-2 is also computationally efficient, generating forecasts for 100 time series in under one second on a single mid-range GPU, 50x and 1000x faster end-to-end than the best task-specific point and probabilistic forecasters, respectively. In addition, we demonstrate its effectiveness in diverse scientific settings -- including infectious-disease forecasting, macroeconomic modeling, carbon-flux prediction, and runoff simulation -- where the model matches or surpasses specialized methods developed for each domain. These results show that a single pretrained model can capture structure shared across temporal systems.
ER -