FrontierAugust 31, 2026via MarkTechPost

Google AI Releases TimesFM-3: A 330M Parameter Zero-Shot Foundation Model For Multivariate Time Series Forecasting

Why it matters

Google ships a 330M parameter foundation model that sets new state-of-the-art on multivariate time series forecasting across three major benchmarks, but the non-commercial license severely limits practitioner adoption and creates a capability-vs.-access tension in the frontier labs.

Key signals

  • 330M parameter model, native multivariate forecasting (no fine-tuning required)
  • Top average rank on GIFT-Eval, fev-bench, and TIME leaderboard
  • Accepts multiple targets, past covariates, past-future covariates in single forward pass
  • Non-commercial, non-production license restricts deployment
  • Advancement over TimesFM 2.5 and prior checkpoints

The hook

Google's TimesFM-3 tops every time-series benchmark—but you can't use it in production.

Google Research has released TimesFM-3, a 330 million parameter time series foundation model that forecasts multiple related series in a single forward pass. Unlike every TimesFM checkpoint through 2.5, it is pretrained natively for multivariate forecasting, accepting multiple targets, past covariat

The week's key stories, every Friday.

For practitioners and enthusiasts — free, in your inbox.

Free forever. No spam.