Google Cloud

TimesFM 3.0 · BigQuery ML

Managed forecasting with BigQuery ML

Open in Studio

What to provide

Enter 12 to 512 evenly spaced, aggregated values and choose 1 to 24 periods. Only use series free of personal or regulated data.

Example input

Series: 12, 15, 17, 16, 19, 18, 20, 24, 22, 25, 26, 29
Horizon: 6

How it works

Studio sends the numeric values to Google Cloud. An AI.FORECAST query returns a point forecast, an 80% interval and a validation forecast for calculating mean absolute error.

How to read the result

TimesFM 3.0 in BigQuery is in preview. Use the same periods and units when comparing forecasts. Actual interval coverage depends on your series.

Credits and licenses

TimesFM 3.0 · Google BigQuery MLGoogle Cloud

Studio uses the managed preview model under the Google Cloud terms. Downloadable weights have a separate license for non-commercial, non-production use.

Official documentation