Open blockchain datasets, built with chainstar and published as Parquet.
Every dataset is documented at bitprimitive.com/data: its columns, row count, the exact blocks it covers, the tools that read them, and a SHA-256 checksum. You can preview each one with SQL in your browser before downloading.
Files are attached to the releases of this repository, one release per build date, each with a SHA256SUMS file.
curl -LO https://github.com/Bit-Primitive/open-data/releases/download/2026-10-21/ethereum-erc20-transfers-sample.parquet
duckdb -c "SELECT token_symbol, count(*) FROM 'ethereum-erc20-transfers-sample.parquet' GROUP BY ALL ORDER BY 2 DESC LIMIT 10"The datasets are licensed under Creative Commons Attribution 4.0 (CC BY 4.0). Cite as: Bit Primitive (year). Dataset name. CC BY 4.0. https://bitprimitive.com/data/<dataset-id>
The underlying blockchain records are public. Token names and symbols are whatever each contract reports, and are not unique: identify tokens by contract address.