Bigframes

BigQuery DataFrames, known as BigFrames, is an open source Python library from Google that lets data scientists analyse and model very large datasets directly inside BigQuery. It offers a pandas compatible API, translates Python code into optimised SQL and runs the work across BigQuery's distributed infrastructure, so datasets far larger than local memory can be processed without downloading them first. It is aimed at analysts, data scientists and machine learning engineers who already work in pandas and want to scale to petabyte sized tables. The bigframes.pandas module covers familiar operations such as read_gbq, groupby, merge and pivot_table for exploration, cleaning and transformation. The bigframes.ml module exposes BigQuery ML algorithms through a familiar machine learning interface, and the bigframes.bigquery.ai submodule brings Gemini models, AI functions and vector search into dataframe workflows. Results can be pulled back into local pandas with to_pandas when needed, and because processing stays inside BigQuery, data remains within existing governance and security controls. It installs with pip and works locally, in Colab or in notebooks attached to a Google Cloud project. The library itself is free and open source, and the documentation site publishes no separate price. The queries and models it runs execute in BigQuery under the user's Google Cloud project, so those costs follow BigQuery's own billing rather than any BigFrames fee.