v2.38.0 (2026-03-16)

Features

• Add bigframes.bigquery.rand() function (#2501) (5c43efb7)

• add df.bigquery pandas accessor (#2513) (91b6c245)

• Add str, dt accessors to pd.col Expression objects (#2488) (ce5de570)

• add bigquery.ml.get_insights function (#2493) (d29a6095)

• use EUC for AI IF, CLASSIFY, and SCORE when connection is not provided (#2507) (fe94910a)

Bug Fixes

• handle unsupported types and empty results in describe (#2506) (2326ad6a)

• no longer automatically use anywidget in the %%bqsql magics (#2504) (43353e2b)

Documentation

• Fix typo in ExperimentOptions class docstring (#2498) (077cb2eb)

• add notebooks to user guide page (#2505) (5cf37888)

v2.37.0 (2026-03-03)

Features

• Add cloud_function_cpus option to remote_function (#2475) (4caf74cc)

• add display.render_mode to control DataFrame/Series visualization (#2413) (7813eaa6)

• Support pd.col expressions with .loc and getitem (#2473) (ae5c8b32)

• add support for Python 3.14 (#2232) (c25a6d01)

• Support pd.col simple aggregates (#2480) (cb00daab)

• add dt.tz_localize() (#2469) (f70f93a1)

• Update bigquery.ai.generate_table output_schema to allow Mapping type (#2463) (f7fd1895)

Bug Fixes

• upload local data through write API if nested JSONs detected (#2478) (01dc5a34)

• allow IsInOp with same dtypes regardless nullable (#2466) (1d81b414)

Documentation

• Add code examples to configuration docstrings (#2352) (3c21993e)

• Move readme content to new User Guide section (#2464) (61a94845)

• add code sample and docstring for bpd.options.experiments.sql_compiler (#2474) (867951bc)

• use direct API for image (#2465) (8a1a82f7)

• Fix recall_score doc example (#2477) (a6f499c1)

• Skip inherited methods, use autosummary only for big classes (#2470) (a9512498)

• add bigframes default connection warning (#2471) (f1bbba23)

2.36.0 (2026-02-17)

Features

• Initial support for biglake iceberg tables (#2409) (ae35a989)

• add bigquery.ai.generate_table function (#2453) (b925aa24)

Documentation

• fix generate_text and generate_table input docs (#2455) (078bd32e)

• update multimodal dataframe notebook to use public APIs (#2456) (342fa723)

• use direct API for pdf chunk and pdf extract (#2452) (543ce52c)

• use direct API for audio transcription (#2447) (59cbc5db)

• Add EXIF metadata extraction example to multimodal notebook (#2429) (84c6f883)

• Update multimodal notebook to use public runtime helpers (#2451) (e36dd8b4)

2.35.0 (2026-02-07)

Features

• Add bigframes.pandas.col with basic operators (#2405) (12741677)

• remove redundant "started." messages from progress output (#2440) (2017cc2f)

• add bigframe.bigquery.load_data function (#2426) (4b0f13b2)

• Disable progress bars in Anywidget mode (#2444) (4e2689a1)

• add bigquery.ai.generate_text function (#2433) (5bd0029a)

• Disable progress bars in Anywidget mode to reduce notebook clutter (#2437) (853240da)

• Add a bigframes cell magic for ipython (#2395) (e6de52de)

• add bigframes.bigquery.ai.generate_embedding (#2343) (e91536c8)

Bug Fixes

• always display the results in the %%bqsql cell magics output (#2439) (2d973b54)

• exlcude gcsfs 2026.2.0 (#2445) (311de31e)

• suppress JSONDtypeWarning in Anywidget mode and clean up progress output (#2441) (e0d185ad)

Documentation

• fix cast method shown on public docs (#2436) (ad0f33c6)

2.34.0 (2026-02-02)

Features

• add bigframes.pandas.options.experiments.sql_compiler for switching the backend compiler (#2417) (7eba6ee03f07938315d99e2aeaf72368c02074cf)
• add bigquery.ml.generate_embedding function (#2422) (35f3f5e6f8c64b47e6e7214034f96f047785e647)
• add bigquery.create_external_table method (#2415) (76db2956e505aec4f1055118ac7ca523facc10ff)
• add deprecation warnings for .blob accessor and read_gbq_object_table (#2408) (7261a4ea5cdab6b30f5bc333501648c60e70be59)
• add bigquery.ml.generate_text function (#2403) (5ac681028624de15e31f0c2ae360b47b2dcf1e8d)

Bug Fixes

• broken job url (#2411) (fcb5bc1761c656e1aec61dbcf96a36d436833b7a)

2.33.0 (2026-01-22)

Features

• add bigquery.ml.transform function (#2394) (1f9ee373)

• Add dark mode to anywidget mode (#2365) (2763b41d)

• Add max_columns control for anywidget mode (#2374) (34b5975f)

• Configure Biome for Consistent Code Style (#2364) (81e27b3d)

• Add BigQuery ObjectRef functions to bigframes.bigquery.obj (#2380) (9c3bbc36)

• Stabilize interactive table height to prevent notebook layout shifts (#2378) (a634e976)

Bug Fixes

• implement retry logic for cloud function endpoint fetching (#2369) (0f593c27)

• Throw if write api commit op has stream_errors (#2385) (7abfef05)

Performance Improvements

• Avoid requery for some result downsample methods (#2219) (95763ff2)

2.32.0 (2026-01-05)

Documentation

• generate sitemap.xml for better search indexing (#2351) (7d2990f1c48c6d74e2af6bee3af87f90189a3d9b)
• update supported pandas APIs documentation links (#2330) (ea71936ce240b2becf21b552d4e41e8ef4418e2d)
• Add time series analysis notebook (#2328) (369f1c0aff29d197b577ec79e401b107985fe969)

Features

• Enable multi-column sorting in anywidget mode (#2360) (1feb956e4762e30276e5b380c0633e6ed7881357)
• display series in anywidget mode (#2346) (7395d418550058c516ad878e13567256f4300a37)
• Refactor TableWidget and to_pandas_batches (#2250) (b8f09015a7c8e6987dc124e6df925d4f6951b1da)
• Auto-plan complex reduction expressions (#2298) (4d5de14ccdd05b1ac8f50c3fe71c35ab9e5150c1)
• Display custom single index column in anywidget mode (#2311) (f27196260743883ed8131d5fd33a335e311177e4)
• add fit_predict method to ml unsupervised models (#2320) (59df7f70a12ef702224ad61e597bd775208dac45)

Bug Fixes

• vendor sqlglot bigquery dialect and remove package dependency (#2354) (b321d72d5eb005b6e9295541a002540f05f72209)
• bigframes.ml fit with eval data in partial mode avoids join on null index (#2355) (7171d21b8c8d5a2d61081f41fa1109b5c9c4bc5f)
• Improve strictness of nan vs None usage (#2326) (481d938fb0b840e17047bc4b57e61af15b976e54)
• Correct DataFrame widget rendering in Colab (#2319) (7f1d3df3839ec58f52e48df088057fc0df967da9)
• Fix pd.timedelta handling in polars comipler with polars 1.36 (#2325) (252644826289d9db7a8548884de880b3a4fccafd)

2.31.0 (2025-12-10)

Features

• add bigframes.bigquery.ml methods (#2300) (719b278c)

• add 'weekday' property to DatatimeMethod (#2304) (fafd7c73)

Bug Fixes

• cache DataFrames to temp tables in bigframes.bigquery.ml methods to avoid time travel (#2318) (d9938319)

Reverts

• DataFrame display uses IPython's _repr_mimebundle_ (#2316) (e4e3ec85)

2.30.0 (2025-12-03)

Features

• Support mixed scalar-analytic expressions (#2239) (20ab469d)

• Allow drop_duplicates over unordered dataframe (#2303) (52665fa5)

• Preserve source names better for more readable sql (#2243) (64995d65)

• use end user credentials for bigframes.bigquery.ai functions when connection_id is not present (#2272) (7c062a68)

• pivot_table supports fill_value arg (#2257) (8f490e68)

• Support builtins funcs for df.agg (#2256) (956a5b00)

• add bigquery.json_keys (#2286) (b487cf1f)

• Add agg/aggregate methods to windows (#2288) (c4cb39dc)

• Add bigframes.pandas.crosstab (#2231) (c62e5535)

• Implement single-column sorting for interactive table widget (#2255) (d1ecc61b)

Bug Fixes

• Pass credentials properly for read api instantiation (#2280) (3e3fe259)

• Update max_instances default to reflect actual value (#2302) (4489687e)

• Improve Anywidget pagination and display for unknown row counts (#2258) (508deae5)

• Fix issue with stream upload batch size upload limit (#2290) (6cdf64b0)

• calling info() on empty dataframes no longer leads to errors (#2267) (95a83f77)

• do not warn with DefaultIndexWarning in partial ordering mode (#2230) (cc2dbae6)

Documentation

• update docs and tests for Gemini 2.5 models (#2279) (08c0c0c8)

• Add Google Analytics configuration to conf.py (#2301) (0b266da1)

• fix LogisticRegression docs rendering (#2295) (32e53134)

• update API reference to new dataframes.bigquery.dev location (#2293) (da064397)

• use autosummary to split documentation pages (#2251) (f7fd2d20)

Full Changelog: v2.29.0...v2.29.1

Bug Fixes

• Bound sqlglot version below 28 (#2275)