Chronos-2

Chronos-2 is a 120M-parameter, encoder-only time series foundation model for zero-shot forecasting that supports univariate, multivariate, and covariate-informed tasks within a single architecture. Inspired by the T5 encoder, it produces multi-step-ahead quantile forecasts and uses a group attention mechanism for efficient in-context learning across related series and covariates. Trained on a combination of real-world and large-scale synthetic datasets, Chronos-2 achieves state-of-the-art zero-shot accuracy among public models on fev-bench, GIFT-Eval, and Chronos Benchmark II. It delivers over 300 time series forecasts per second on a single A10G GPU and supports both GPU and CPU inference. Chronos-2 achieves a win rate of over 90% against Chronos-Bolt in head-to-head comparisons. It was released on October 20, 2025, under the Apache-2.0 license.

Capabilities

Capability	Chronos-2
Univariate Forecasting	✅
Cross-learning across items	✅
Multivariate Forecasting	✅
Past-only (real/categorical) covariates	✅
Known future (real/categorical) covariates	✅
Max. Context Length	8192
Max. Prediction Length	1024

Benchmarks and Evaluation

Chronos-2 delivers state-of-the-art performance across three comprehensive benchmarks: fev-bench, GIFT-Eval, and Chronos Benchmark II. On fev-bench, which emphasizes multivariate and covariate-informed forecasting, its universal in-context learning capabilities lead to substantial improvements over existing models. On tasks involving covariates, it consistently outperforms baselines by a wide margin. Case studies in the energy and retail domains further highlight its practical advantages.

Additional Resources

• Technical report
• GitHub repository
• Amazon Science Blog