# amazon-science/chronos-forecasting

Chronos: Pretrained Models for Time Series Forecasting

Repository: https://github.com/amazon-science/chronos-forecasting
Canonical: https://ross.abutalabs.com/products/chronos-forecasting
Homepage: https://arxiv.org/abs/2510.15821
Language: Python
License: Apache-2.0
License Family: permissive
Topics: forecasting, large-language-models, llm, machine-learning, time-series, foundation-models, pretrained-models, time-series-forecasting, timeseries, artificial-intelligence, huggingface, huggingface-transformers, transformers
Last push: 2026-08-14T10:43:25+00:00

## Health v2 (maintenance only)
Score: 88/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 97, release rhythm 91, longevity 65
- inputs: {"age_days": 922, "days_push": 19, "days_rel": 62, "gap_med": 16.5, "n_releases_24m": 15}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 5759, forks 700 (observed 2026-08-28T04:09:29.329676+00:00)

## What it is
Chronos is a Python library providing pretrained foundation models for time series forecasting, including Chronos-2 which handles univariate, multivariate, and covariate-informed tasks zero-shot. Models are distributed via Hugging Face and deployable to AWS SageMaker and AutoGluon-Cloud.

## Use cases
- forecast future values of a time series without training a model
- zero-shot multivariate forecasting with covariates
- generate probabilistic demand or sales forecasts
- deploy a pretrained forecasting model to SageMaker
- benchmark time series forecasting models
- integrate forecasting into pandas workflows

## When to choose
- you need accurate forecasts without task-specific training
- you want state-of-the-art zero-shot performance on multivariate or covariate-informed tasks
- you want fast, memory-efficient inference via Chronos-Bolt models
- you prefer inference-only foundation models over per-dataset model fitting

## When to avoid
- you need interpretable classical statistical methods like ARIMA or ETS
- you have very limited compute and cannot run neural models
- you need online learning or continual model updates
- your data is not time series

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, llm-inference, sdk
- domain: time-series, machine-learning, artificial-intelligence, large-language-models
- platform: python, cloud
- tags: time-series-forecasting, foundation-models, pretrained-models, zero-shot-forecasting, huggingface-transformers, chronos, gpu

## Member repositories
- amazon-science/chronos-forecasting (main) score 88

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:29.329676+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:53:15.091123+00:00, confidence not recorded.
  - readme: https://github.com/amazon-science/chronos-forecasting (fetched 2026-08-28T04:09:29.329676+00:00, sha 879e56a2d313)
  - homepage: https://arxiv.org/abs/2510.15821 (fetched 2026-08-29T08:48:27.097316+00:00, sha be208cf6ce8a)
  - site_page: https://info.arxiv.org/about/donate.html (fetched 2026-08-29T08:48:27.106924+00:00, sha cca9c3a11c56)
  - site_page: https://info.arxiv.org/about/ourmembers.html (fetched 2026-08-29T08:48:27.110446+00:00, sha 47cbc55ff1de)
  - site_page: https://info.arxiv.org/about (fetched 2026-08-29T08:48:27.112221+00:00, sha a1f16f915a9a)
  - registry_pypi: https://pypi.org/pypi/chronos-forecasting/json (fetched 2026-08-29T08:48:27.113835+00:00, sha 1bd2cffa0f9b)
  - site_page: https://info.arxiv.org/labs/index.html (fetched 2026-08-29T08:48:27.108884+00:00, sha b14a8d05a0ec)
- Data as of 2026-08-30T08:39:29.467469+00:00.