Formats

Datasets support the following formats:

• CSV (csvgz)
• MMDB (mmdb)
• JSON (jsonl / ndjson) (Coming Soon)
• Parquet (parquet) (Coming Soon)

Note: Datasets that don't have an IP-based key will not be available in mmdb format, since mmdb is keyed by IP.

To query the databases in Google BigQuery rather than download them, see BigQuery.

CSV Format (csvgz)

csvgz is a CSV file compressed using the gzip (GNU zip) algorithm.

Column names are always included as the first row.

Cells where a value is NULL or the empty string are represented without any value. For example, here the provider is NULL and represented without any cell value:

start_ip,end_ip,provider,confidence,last_seen
0.0.0.0,0.0.0.0,,low,2026-01-01

See how to decompress the CSV file.

MMDB Format (mmdb)

mmdb is the MMDB file format that allows for a fully in-memory O(1) IP lookup database.

The MMDB files are generally much larger than their CSV equivalents, but can be used for serving API requests at very high throughputs and scale.

The following tools are available to work with MMDBs:

• https://github.com/maxmind/mmdbconvert
• https://github.com/maxmind/mmdbinspect
• https://github.com/internetdata/mmdb

JSON Format (jsonl / ndjson)

The JSON files are json-lines (jsonl) and newline-delimited (ndjson) format files compressed using the gzip (GNU zip) algorithm.

Not published yet: a download takes csvgz or mmdb today, and answers 400 with INVALID_FORMAT for jsonl or ndjson.

Each line in the file is an individual self-describing JSON object. For example:

{"start_ip": "0.0.0.0", "end_ip": "0.0.0.0", "provider": "", "confidence": "low", "last_seen": "2026-01-01"}

See how to decompress the JSON file.

Parquet

Coming soon.

Decompression

CSV & JSON files are always compressed using the gzip (GNU zip) algorithm. Decompression is necessary to properly read the files.

They can be decompressed with any tool that supports decompression of the gzip algorithm. Here are some examples for a Linux terminal environment with different utilities:

# Using `gunzip`
gunzip file.csv.gz
gunzip file.jsonl.gz

# Using `gzip`
gzip -d file.csv.gz
gzip -d file.jsonl.gz

# Using `zcat`
zcat file.csv.gz > file.csv
zcat file.jsonl.gz > file.jsonl