Chronos-2
Model directoryamazon/chronos-2120M params | Upstream checkpoint
Chronos-2 is Amazon's universal successor to the original Chronos family, at 120M parameters. The official card frames it as a single model that handles univariate, multivariate, and covariate-informed forecasting in one architecture, a more general design than the univariate-first checkpoints of the first Chronos generation.
Under the hood it is a T5-inspired, encoder-only model that uses group attention and multi-step quantile decoding, so it can reason across related series and native covariates while emitting probabilistic forecasts directly. It is pretrained on a subset of Chronos Datasets, a subset of GIFT-Eval Pretrain, and synthetic univariate and multivariate data, as described by the official Chronos-2 card and technical report.
Where to run this model
Compute availability, hardware requirements, and pricing depend on the execution provider.
Get model weights, deployment instructions, and current provider listings.
- Amazon SageMaker JumpStart
managed deployment · Source checked 2026-10-06
AWS lists Chronos 2 as pytorch-forecasting-chronos-2. Deploy an endpoint in your AWS account and verify its revision and request contract.
- Birla AI Labs Forecasting API
provider listing · Source checked 2026-10-06
The provider lists this model family. Check its current checkpoint, size, license, pricing and input contract before use.
- Retrocast · The Forecasting Company
managed service · Source checked 2026-10-06
Forecast through the managed API using chronos-2. The provider controls the deployed variant and request limits.
- Community forecasting playground
hosted demo · Source checked 2026-10-06
The published loader names this checkpoint. Space was sleeping when checked; it may require a wake-up. Hosted computation, not browser inference.
- sktime TServe
self-hosted runtime · Family / provider variant · Source checked 2026-10-06
Run a model-serving container on your own hardware or compute account. Select the catalog checkpoint and matching runtime image.
Hugging Face provider mappings checked 2026-10-06. No Inference Provider mapping was listed for this checkpoint.
Model Classification
Family
Chronos-2
Type
time series foundation model
Pretrained time-series model for zero-shot or few-shot forecasting workloads.
Resources
Training Data
Subset of Chronos Datasets, subset of GIFT-Eval Pretrain, plus synthetic univariate and multivariate data as described by the official Chronos-2 card and technical report.
Recommended For
- • Zero-shot forecasting with covariates or related multivariate series
- • Teams that want one Chronos model for broader forecasting workloads
Strengths
- • Handles univariate, multivariate, and covariate-aware forecasting in one family
- • More universal design than the original first-generation Chronos checkpoints
Limitations
- • Still a forecasting model, not a general multi-task representation backbone
- • Larger deployment footprint than the smallest lightweight baselines
Capabilities
Tags
Specifications
- Parameters
- 120M
- Architecture
- encoder-only, T5-inspired model with group attention and multi-step quantile decoding
- Context length
- n/a
- Max context
- n/a
- Minimum history
- n/a
- Recommended history
- n/a
- Input step
- n/a
- Required target series
- 1
- Temperature
- Ignored
- Top P
- Ignored
- Max output
- n/a
- License
- n/a