TimesFM 3.0

On the New York subway data, TimesFM 3.0's confidence intervals are too narrow, and before going to production you should check via backtesting whether the frequency of actual values falling within the intervals matches expectations.

On the New York subway data, TimesFM 2.5 made slightly better use of additional future inputs (such as a holiday calendar) than 3.0 and had more reliable uncertainty intervals, while 3.0's inference speed was several times faster.

In the Lumen business simulation, TimesFM 3.0 completed a revenue forecast for the next two months in under a second with a deviation within 2%; when re-evaluated a quarter later, the deviation was still within 6%, and neither version exceeded its respective uncertainty interval.

In a what-if scenario where the marketing budget was doubled, TimesFM 3.0 showed almost no reaction while 2.5 predicted a large jump in revenue; the agent trusted 3.0 and recommended Option A, but the real outcome was that Option B brought in more revenue, and the agent chose wrong in two of three test rounds.

TimesFM 3.0 is a decoder-only Transformer trained entirely on time series data and supports zero-shot forecasting.

TimesFM 3.0 builds covariate handling into the Transformer itself, using attention mechanisms to let each token directly see future covariate inputs, and generates the entire forecast window in one pass to improve inference speed.

Google claims in its paper that TimesFM 3.0 ranks first among foundation models on three public forecasting benchmarks.

In a test on daily New York subway ridership, TimesFM 2.5 and 3.0 reduced forecast error by about 25% relative to a seasonal baseline under zero-shot conditions.

The downloadable weights of TimesFM 3.0 are under a non-commercial license, and commercial use requires BigQuery ML.

According to Google, TimesFM 3.0 leads on the fev-bench, TIME, and GIFT-Eval benchmarks.

TimesFM 3.0 is a foundation model dedicated to time series forecasting, with zero-shot forecasting capability.