GenAIWiki
Models

Time-series foundation model

A time-series foundation model is pretrained on many series so it can forecast new series with little or no task-specific training.

Expanded definition

Unlike a text LLM, a time-series foundation model consumes numeric histories (and sometimes covariates) and outputs future values or quantiles. Google TimesFM-3 is a current example in this catalog: a 330M multivariate forecaster with a non-commercial weight license. These models are not drop-in chat models. You still need to check frequency, missing data, license, and whether the hosted forecast API matches the research checkpoint.

Related terms

Explore adjacent ideas in the knowledge graph.

Time-series foundation model FAQ

What is Time-series foundation model?

A time-series foundation model is pretrained on many series so it can forecast new series with little or no task-specific training.

How is Time-series foundation model used in AI systems?

Unlike a text LLM, a time-series foundation model consumes numeric histories (and sometimes covariates) and outputs future values or quantiles. Google TimesFM-3 is a current example in this catalog: a 330M multivariate forecaster with a non-commercial weight license. These models are not drop-in chat models. You still need to check frequency, missing data, license, and whether the hosted forecast...

Related

Comparisons, tools, and models that connect to this idea.