Google's TimesFM 3.0 Claims Top Spot Across All Major Forecasting Benchmarks With New Multivariate Capabilities

Google Research's TimesFM 3.0 has claimed the #1 ranking across all three major forecasting benchmarks, introducing powerful multivariate capabilities and zero-shot generalization, though its latest pretrained weights come with a non-commercial license restricting production use.

• Google Research releases TimesFM 3.0, a pretrained time-series foundation model that achieves rank #1 across all three major forecasting benchmarks: fev-bench, TIME Benchmark, and GIFT-Eval.
• TimesFM 3.0 introduces native multivariate forecasting, flexible covariate support for both past-only and past-and-future inputs, and superior zero-shot generalist capabilities without per-task tuning.
• The 3.0 pretrained weights are released under a non-commercial license restricting production use, while source code and weights up to version 2.5 remain under the open Apache-2.0 license.