Chronos-Bolt (Small)
Model directoryamazon/chronos-bolt-small48M params | Upstream checkpoint
Chronos-Bolt Small is Amazon's fast zero-shot forecasting checkpoint built for production use, at 48M parameters. It is positioned as a strong default in the Chronos-Bolt family: probabilistic forecasts with low latency and a simpler deployment footprint than the original Chronos line.
It uses a patch-based T5 encoder-decoder and replaces autoregressive token-by-token rollout with direct multi-step quantile prediction over patchified history. Forecasting the entire horizon in one pass is what makes Bolt materially faster and lighter than the first Chronos generation. It is pretrained on nearly 100B time-series observations drawn from large public corpora and synthetic pretraining data, as documented by the official Chronos-Bolt model cards.
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.
- 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.
- Amazon SageMaker JumpStart
managed deployment · Source checked 2026-10-06
AWS lists autogluon-forecasting-chronos-bolt-small. Deploy the AutoGluon-wrapped model in your AWS account.
Hugging Face provider mappings checked 2026-10-06. No Inference Provider mapping was listed for this checkpoint.
Model Classification
Family
Chronos-Bolt
Type
time series foundation model
Pretrained time-series model for zero-shot or few-shot forecasting workloads.
Resources
Training Data
Nearly 100B time-series observations from large public corpora and synthetic pretraining data, as documented by the official Chronos-Bolt model cards.
Recommended For
- • Fast zero-shot probabilistic forecasting in production APIs
- • Teams replacing classic Chronos-T5 with lower latency inference
Strengths
- • Direct multi-step forecasting avoids autoregressive rollout cost
- • Strong default for quantile forecasts with lightweight serving
Limitations
- • Focused on forecasting rather than broader multi-task time-series work
- • Less natural fit when you need model behavior tuned around rich multivariate covariates
Capabilities
Tags
Specifications
- Parameters
- 48M
- Architecture
- patch-based T5 encoder-decoder with direct multi-step quantile forecasting
- 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