Timestral-123M

Timestral is a family of time series foundation models for zero-shot forecasting, available in 5M, 28M, 123M, and 209M parameter versions. This repository contains the 123M checkpoint.

[!Note] For more information about our Timestral family, go to the main model page: Timestral-209M.

Model Details

• Size: 123M parameters.
• Architecture: patch-based transformer with time attention and variate attention, 20 layers, embedding dimension 640.
• Patching: 32 time steps per patch. Each patch is embedded from its values together with their missing-value indicators.
• Context length: up to 32,768 time steps (1,024 patches).
• Forecast horizon: Up to 1,024 time steps (32 patches) in a single forward pass. Longer horizons are forecast block by block, with each block's median forecast appended to the context.
• Outputs: 21 quantiles (0.01, 0.05, 0.1, 0.15, …, 0.85, 0.9, 0.95, 0.99).
• Inputs: univariate or multivariate target variates, plus past-only and known-future covariates, numeric or categorical (categorical covariates are target-encoded). Missing values are marked as NaN.
• Inference: PyTorch on CPU or GPU.

Benchmarks

Timestral-209M achieves state-of-the-art zero-shot accuracy across three major forecasting benchmarks: GIFT-Eval, TIME, and fev-bench.

Smaller Timestral models offer a trade-off between model size and forecasting performance. All Timestreal models outperform every other model at the respective model size. See the Pareto-front plot on the main model page.

Results across the Timestral family

License

Models have been released under a Modified MIT license: See here.

You must not use this model in a manner that infringes, misappropriates, or otherwise violates any third party’s rights, including intellectual property rights