@ForecastEng: Amazon quietly open-sourced a time-series model that forecasts out of the box. No training. No feature engineering. Poi…
Summary
Amazon open-sourced Chronos, a time-series forecasting model that predicts out of the box without training or feature engineering, treating forecasting like language models treat text.
View Cached Full Text
Cached at: 06/24/26, 10:22 AM
Amazon quietly open-sourced a time-series model that forecasts out of the box.
No training. No feature engineering. Point it at your data and it predicts.
It’s called Chronos — and it treats forecasting the way a language model treats text.
Let me break it down
Chronos-2 is a pretrained model capable of handling univariate, multivariate, and covariate-informed forecasting tasks in a zero-shot manner
Chronos 2 implements a pipeline that internally normalizes the time series, tokenizes, and produces forecasts using the architecture below.
Forecasting with Chronos 2 in 7 lines of code:
Whitepaper (31 page pdf): https://arxiv.org/pdf/2510.15821
Github: https://github.com/amazon-science/chronos-forecasting…
Companies don’t want another ARIMA model.
They want engineers who can build forecasting systems that actually ship to production.
Wednesday, July 8th — I’m sharing mine live in Python.
Free. Full code. 670+ already registered.
Save your seat → https://learn.business-science.io/forecast-register…
That’s a wrap!
If you enjoyed this thread:
- Follow me @ForecastEng for more of these
- RT the tweet below to share this thread with your audience
Similar Articles
@nicos_ai: GOOGLE HAS SILENTLY RELEASED AN AI THAT PREDICTS PATTERNS Sales. Market prices. Web traffic. Energy demand. Crypto vola…
Google has released TimesFM, an AI model for zero-shot time series forecasting, trained on 100 billion real data points, free and open-source.
Unified Zero-Shot Time Series Forecasting: A Darts Foundation
Darts, a popular open-source Python library for time series analysis, introduces a unified FoundationModel class collection that integrates multiple time series foundation models (Chronos-2, TimesFM 2.5, TiRex, PatchTST-FM) for zero-shot and fine-tuned forecasting with standardized interfaces and minimal dependencies.
TimesFM-3: A zero-shot foundation model for multivariate forecasting
Google introduces TimesFM-3, a state-of-the-art zero-shot foundation model for multivariate time series forecasting, capable of handling multiple targets and covariates in a single forward pass without fine-tuning.
google-research/timesfm
Google Research releases TimesFM 2.5, an open-source time-series foundation model for forecasting, with 200M parameters, 16k context length, and support for quantile forecasts up to 1k horizon. The model is available on PyPI and Hugging Face, with fine-tuning via LoRA and integration into Google products like BigQuery ML, Google Sheets, and Vertex Model Garden.
@GoogleDeepMind: We’re open sourcing the code and model weights on @Github, making them freely available for anyone to build on. This co…
Google DeepMind is open sourcing WeatherNext, an AI model that achieves state-of-the-art accuracy in cyclone forecasting, offering an extra day of warning and representing a decade of meteorological progress.