Google Research releases TimesFM 3.0 forecasting model
Google Research released TimesFM 3.0 with native univariate and multivariate time-series forecasting and support for past-only and past-and-future dynamic covariates. A PyTorch checkpoint is available as google/timesfm-3.0-pytorch, while the repository also describes an MLX backend for Apple silicon. The project claims leading results on three time-series benchmarks, but those claims are not independently verified in the supplied report.
TimesFM 3.0 expands Google Research’s time-series foundation model with multivariate forecasting and covariate support, although its downloadable weights are limited to non-commercial, non-production use.
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.

https://research.google/blog/a-decoder-only-foundation-model-for-time-series-forecasting/
TimesFM
TimesFM (Time Series Foundation Model) is a pretrained time-series foundation model developed by Google Research for time-series forecasting.