Published on July 23, 2026, the announcement describes a dataset with about 5 trillion tokens across 713 languages and 224 million repositories, collected through August 2025.
The July 23, 2026 announcement presents The Stack v3, a dataset with about 5 trillion source-code tokens across 713 filtered languages and 224 million repositories. The GitHub collection underlying the dataset was completed through August 2025.
According to the description, engineers can use the code as a source for training or fine-tuning models. To check the scope and details, consult the original announcement and verify its coverage, filters, collection date, and dataset terms of use.