--- title: Google brings predictive AI to BigQuery without the ML... description: Google is adding TabFM, a pre-trained foundation model for tabular data, to BigQuery in preview, letting teams run classification and regression predictions... canonical: https://daily.dev/posts/google-brings-predictive-ai-to-bigquery-without-the-ml-training-u6h0nyfhl twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Google brings predictive AI to BigQuery without the ML training | daily.dev og:description: Google is adding TabFM, a pre-trained foundation model for tabular data, to BigQuery in preview, letting teams run classification and regression predictions... og:url: https://daily.dev/posts/google-brings-predictive-ai-to-bigquery-without-the-ml-training-u6h0nyfhl og:image: https://api.daily.dev/og/posts/U6H0Nyfhl.png og:image:alt: Google brings predictive AI to BigQuery without the ML training og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Google brings predictive AI to BigQuery without the ML training **[InfoWorld](https://daily.dev/sources/infoworld)** · 4 min read · 0 upvotes · 0 comments ## Summary Google is adding TabFM, a pre-trained foundation model for tabular data, to BigQuery in preview, letting teams run classification and regression predictions using historical data as in-context examples via the new AI.PREDICT and AI.EVALUATE SQL functions, without training or deploying a separate ML model. This could cut engineering and infrastructure overhead and let SQL users generate predictions directly. However, TabFM currently caps at 20 feature columns, Google still recommends XGBoost or custom models for very large datasets or when feature-importance explainability is needed, and Google plans to move TabFM to token-based pricing on top of standard BigQuery charges starting October 30, 2026, which analysts warn could raise costs for high-frequency, large-scale prediction workloads. Analysts advise piloting it on real workloads before shifting ML budgets. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is TabFM in BigQuery and how does it let me make predictions without training a model? TabFM is a pre-trained foundation model for tabular data, developed by Google Research, now available in preview in BigQuery. It performs classification and regression by using labeled historical data as in-context examples rather than requiring a trained model, accessed through the new AI.PREDICT SQL function, with AI.EVALUATE available to measure prediction accuracy against known outcomes. _Teams weighing TabFM against traditional ML pipelines can track real-world BigQuery workload comparisons on daily.dev._ ### When does BigQuery's TabFM model switch to token-based pricing? Google plans to move TabFM to token-based pricing on top of standard BigQuery charges starting October 30, 2026. Analysts warn this could make TabFM more expensive than a trained and cached model for large-scale or high-frequency prediction workloads, so enterprises should pilot the model on their own query volumes before committing significant ML budget to it. _Anyone budgeting BigQuery ML costs can follow pricing shifts like this one via daily.dev._ ### Should I use TabFM or XGBoost for tabular data predictions in BigQuery? Use XGBoost or a custom model for very large historical datasets, datasets exceeding TabFM's current limit of 20 feature columns, cases needing full control over hyperparameter tuning, or when feature-importance explainability is required. TabFM suits smaller, lower-frequency or experimental workloads where avoiding a training pipeline saves time and infrastructure cost. _Developers deciding between TabFM and XGBoost for tabular predictions can compare tradeoffs on daily.dev._ ## Similar posts on daily.dev - [TimesFM models in BigQuery and AlloyDB](https://daily.dev/posts/timesfm-models-in-bigquery-and-alloydb-haojhbaki) · Google Cloud · 0 upvotes · 0 comments - [How the Rise of Tabular Foundation Models Is Reshaping Data Science](https://daily.dev/posts/how-the-rise-of-tabular-foundation-models-is-reshaping-data-science-2taes3vse) · Towards Data Science · 0 upvotes · 0 comments - [AI-based forecasting and analytics in BigQuery via MCP and ADK](https://daily.dev/posts/ai-based-forecasting-and-analytics-in-bigquery-via-mcp-and-adk-mldwfndnv) · Google Cloud · 1 upvotes · 0 comments --- Tags: [#machine-learning](https://daily.dev/tags/machine-learning), [#gcp](https://daily.dev/tags/gcp), [#data-analysis](https://daily.dev/tags/data-analysis), [#google-bigquery](https://daily.dev/tags/google-bigquery), [#xgboost](https://daily.dev/tags/xgboost) [View this post on daily.dev](https://daily.dev/posts/google-brings-predictive-ai-to-bigquery-without-the-ml-training-u6h0nyfhl) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Google brings predictive AI to BigQuery without the ML training","url":"https://daily.dev/posts/google-brings-predictive-ai-to-bigquery-without-the-ml-training-u6h0nyfhl","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/google-brings-predictive-ai-to-bigquery-without-the-ml-training-u6h0nyfhl"},"datePublished":"2026-09-02T12:18:51.240Z","dateModified":"2026-09-02T15:46:08.111Z","description":"Google is adding TabFM, a pre-trained foundation model for tabular data, to BigQuery in preview, letting teams run classification and regression predictions...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1c5c3b3a6cbeb9aaea1c0477abf70367?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1c5c3b3a6cbeb9aaea1c0477abf70367?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"InfoWorld","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"InfoWorld","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/bf6d68a999064029b0bb09aa6268f1f3","url":"https://daily.dev/sources/infoworld"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/google-brings-predictive-ai-to-bigquery-without-the-ml-training-u6h0nyfhl","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"machine-learning,gcp,data-analysis,google-bigquery,xgboost","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"InfoWorld","item":"https://daily.dev/sources/infoworld"},{"@type":"ListItem","position":3,"name":"Google brings predictive AI to BigQuery without the ML training"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/google-brings-predictive-ai-to-bigquery-without-the-ml-training-u6h0nyfhl#faq","mainEntity":[{"@type":"Question","name":"What is TabFM in BigQuery and how does it let me make predictions without training a model?","acceptedAnswer":{"@type":"Answer","text":"TabFM is a pre-trained foundation model for tabular data, developed by Google Research, now available in preview in BigQuery. It performs classification and regression by using labeled historical data as in-context examples rather than requiring a trained model, accessed through the new AI.PREDICT SQL function, with AI.EVALUATE available to measure prediction accuracy against known outcomes. Teams weighing TabFM against traditional ML pipelines can track real-world BigQuery workload comparisons on daily.dev."}},{"@type":"Question","name":"When does BigQuery's TabFM model switch to token-based pricing?","acceptedAnswer":{"@type":"Answer","text":"Google plans to move TabFM to token-based pricing on top of standard BigQuery charges starting October 30, 2026. Analysts warn this could make TabFM more expensive than a trained and cached model for large-scale or high-frequency prediction workloads, so enterprises should pilot the model on their own query volumes before committing significant ML budget to it. Anyone budgeting BigQuery ML costs can follow pricing shifts like this one via daily.dev."}},{"@type":"Question","name":"Should I use TabFM or XGBoost for tabular data predictions in BigQuery?","acceptedAnswer":{"@type":"Answer","text":"Use XGBoost or a custom model for very large historical datasets, datasets exceeding TabFM's current limit of 20 feature columns, cases needing full control over hyperparameter tuning, or when feature-importance explainability is required. TabFM suits smaller, lower-frequency or experimental workloads where avoiding a training pipeline saves time and infrastructure cost. Developers deciding between TabFM and XGBoost for tabular predictions can compare tradeoffs on daily.dev."}}]} ```