Open data
Veil's three datasets, free and citable.
Persian tokenizer efficiency
Exact token counts for a fixed paired Persian/English corpus, across every model tokenizer that is publicly resolvable. Veil's own measurement, reproducible from the harness and the corpus, both published here. Not an estimate.
Iran AI Access Matrix
Field-report summaries per provider — aggregates only. No raw report is ever published, and a cell backed by fewer than five distinct reporters exports as null rather than as a small number.
Exported weekly to the public mirror. Ask if you need a snapshot before the next run.
Bilingual lexicon
AI technical vocabulary defined in Persian and English.
Exported weekly to the public mirror. Ask if you need a snapshot before the next run.
Entity specifications
Every fact veil publishes, with its source URL and date.
Exported weekly to the public mirror. Ask if you need a snapshot before the next run.