v2026.8.0 (2026-08-07)

This is a regular quarterly PUDL release with a bunch of dataset updates. We've also added a new raw EPA MATS dataset, and Puerto Rico is now included in our EIA-860/860M data thanks to first-time contributor @bsousa22. We also tracked down and fixed a longstanding bug that had FERC Form 714 balancing authority data crossed between the Desert Southwest and Upper Great Plains West regions. See below for all the details.

New Data

EIA-860M

• Added Puerto Rico EIA-860M data into EIA 860 tables. See issue #4352 and PR #5360. Shoutout to @bsousa22 for making his first PUDL contribution!

EPA MATS

• Added the EPA Mercury and Air Toxics Standards (MATS) dataset, extracted into a raw table with additional records through end of June 2026. This data is not yet deeply integrated into PUDL. See issue #5358 and PRs #5389, #5464. Also a contribution from @bsousa22.

Expanded Data Coverage

EIA-191

• Updated EIA-191 data. See PR #5464.

EIA-860M

• Added EIA-860M data through June 2026. See issue #5459 and PR #5468.

EIA-923

• Added early release data for EIA-923 2025. See issue #5372 and PR #5391.

• Added EIA-923M data for April and May 2026. See issue #5460 and PRs #5391, #5468.

EIA-930

• Updated EIA-930 data. See PRs #5445, #5464.

FERC-714

• Added 2025 XBRL data for FERC-714. See #5424 and PRs #5436, #5464.

EIA Electricity API

• Updated the bulk EIA Electricity API data used to fill in redacted fuel prices. See PRs #5441, #5464.

EPA CEMS

• Updated the EPA CEMS data with additional records through end of June 2026. See PRs #5441, #5464.

PHMSA Natural Gas data

• Updated the PHMSA natural gas data. See PR #5464.

FERC Form 6

• Updated the raw FERC Form 6 archives to include additional 2025 data. This data is converted to SQLite, but not deeply integrated into PUDL. See PR #5441.

Documentation

• Added LLM use guidelines and best practices to the contributor guide and dev guide. See PR #5395.

• Set up the sphinx_llm Sphinx extension to generate a Markdown version of the PUDL documentation, suitable for consumption by LLMs, based on the llms.txt convention. Each page now advertises its Markdown counterpart via a <link rel="alternate" type="text/markdown"> tag, and the site footer links directly to llms.txt, so that agents browsing the rendered HTML docs can discover and prefer the Markdown versions. See PRs #5381, #5393.

Bug Fixes & Data Cleaning

• Fixed incorrectly mapped Western Area Power Authority BA codes in FERC 714 data - previously, the Upper Great Plains West region FERC respondent was mapped to the Desert Southwest region EIA balancing authority information, and vice versa. See #4644 and #5408.

• Fixed an exact-float merge bug in FERC1 "exploded tables" corrections. The add_sizable_minority_corrections() method matched correction candidates by merging directly on floating point columns. Exact float equality is extremely brittle, and moving from 32- to 64-bit floats changed which utility/year pairs matched, producing extra rows and different ending_balance sums in out_ferc1__yearly_detailed_balance_sheet_assets and out_ferc1__yearly_rate_base. These were real matches being lost due to the limited precision of 32-bit floats, on top of the brittleness of of merging on floating point numbers. These are now being captured deterministically by merging on values within a fixed tolerance of $5. See PR #5350.

• Fixed the Dagster asset graph for EIA-860 to correctly depend on EIA-860M, so selecting the raw_eia860m asset group and its downstream assets now also captures the EIA-860 assets that rely on it, instead of silently leaving them stale. See #4327 and PR #5409.

Performance Improvements

• Switched from using pl.Enum to unconstrained pl.Categorical types to preserve lazy execution in get_parquet_table_polars(). This allows running schema checks on large tables, and Pandera's Polars backend never actually enforced data content checks on pl.LazyFrame assets (99% of assets in PUDL). See PR #5434. Implementing efficient content validation for pl.LazyFrame assets is left to PR #5432.

• The fast ETL now processes only two representative EIA-861 years instead of the entire time series, bringing it in line with how every other dataset is already handled and speeding up both local development and CI. Processing all years was originally a workaround for discontinued columns and data validation tests that couldn't tolerate partial coverage; those limitations have since been resolved. This change surfaced an implicit assumption in the FERC-714 outputs that all EIA-861 years were always available, in the logic that repairs known-bad balancing authority/utility associations by copying data from a known-good year. That repair logic has been rewritten as an explicit, validated mapping of per-year fixes, so it degrades gracefully when only a subset of years is present, and is substantially easier to read, test, and extend than the compact form it replaces. See #2628 and #4568.

Developer Experience

• Standardized numeric dtypes to always use 64-bit values, and datetime types to use microsecond resolution in pudl.metadata.dtypes. The unused compact argument to to_pandas_dtype() was retired. See PR #5350.

• Sanitized the batch job IDs used by the deploy and build workflows so they always meet Google Batch's naming requirements, fixing failures we had previously worked around by making the IDs more generic. See #5411 and PR #5429.

Other PUDL v2026.8.0 Resources

• PUDL v2026.8.0 Data Dictionary
• PUDL v2026.8.0 Documentation
• PUDL in the AWS Open Data Registry
• PUDL v2026.8.0 in a free, public AWS S3 bucket: s3://pudl.catalyst.coop/v2026.8.0/
• PUDL v2026.8.0 in a requester-pays GCS bucket: gs://pudl.catalyst.coop/v2026.8.0/
• PUDL v2026.8.0 release on GitHub

Contact Us

If you're using PUDL, we would love to hear from you! Even if it's just a note to let us know that you exist, and how you're using the software or data. Here's a bunch of different ways to get in touch:

• Follow us on GitHub
• Use the PUDL Github issue tracker to let us know about any bugs or data issues you encounter
• GitHub Discussions is where we provide user support.
• Watch our GitHub Project to see what we're working on.
• Email us at hello@catalyst.coop for private communications.
• On Mastodon: @CatalystCoop@mastodon.energy
• On BlueSky: @catalyst.coop
• Connect with us on LinkedIn
• Play with our data and notebooks on Kaggle
• Combine our data with ML models on HuggingFace
• Learn more about us on our website: https://catalyst.coop
• Subscribe to our announcements list for email updates.