Google Cloud announced the preview of TimesFM 3.0 in BigQuery on October 12, 2026, introducing native multivariate time-series forecasting. Developed by Google Research, the 330-million parameter model brings advanced forecasting capabilities directly into SQL workflows.
Multivariate Foundation Modeling
Google Cloud said TimesFM 3.0 is a 330-million parameter foundation model developed by Google Research. According to the company, the model ranks first across the GIFT-Eval, FEV-Bench, and TIME public time-series leaderboards.
The update enables users to perform multivariate time-series forecasting, handle past-only covariates, and utilize future covariates using SQL functions like AI.FORECAST.
Integration and Conversational Analytics
Google Cloud stated that TimesFM 3.0 serves as a drop-in upgrade for univariate forecasting and anomaly detection via AI.DETECT_ANOMALIES. Existing workflows can utilize the model by specifying model => 'TimesFM 3.0'.
Users can also run multivariate forecasts using natural language in BigQuery Conversational Analytics by asking prompts for specific dates and covariates.
What to do
- Try AI.FORECAST with model => 'TimesFM 3.0' directly in the BigQuery Studio SQL workspace.
- Try AI.DETECT_ANOMALIES with model => 'TimesFM 3.0' for univariate time series anomaly detection.
- Read the Google Research blog post and explore the BigQuery TimesFM documentation.
Key facts and where they come from
- Google Cloud announced the preview of TimesFM 3.0 in BigQuery on October 12, 2026.
Today, we are thrilled to announce the preview of TimesFM 3.0 in BigQuery.
- TimesFM 3.0 is a 330-million parameter model developed by Google Research.
Developed by Google Research, TimesFM 3.0 is a 330-million parameter state-of-the-art foundation model
- The model ranks #1 across GIFT-Eval, FEV-Bench, and TIME leaderboards.
TimesFM 3.0 ranks #1 across all three major public time-series leaderboards — GIFT-Eval, FEV-Bench, and TIME
