Google adds multivariate forecasting to BigQuery AI

Google just launched TimesFM 3.0 for BigQuery, letting you run multivariate time-series forecasts directly in SQL. Unlike earlier versions that handled only single outcomes, TimesFM 3.0 can predict multiple, related future values—like demand shaped by weather, holidays, or pricing—all at once. Google says it now leads all major public forecasting benchmarks. You don’t have to retrain models or leave your workflow—just specify the new model.

• TimesFM 3.0 supports multivariate time-series forecasting
• Available directly in BigQuery SQL workflows
• Claims top scores on three leading public benchmarks
• No need to retrain or change existing workflows