Veil
The system of record for the AI industry, in Persian and English
Every fact here is one claim, with the address it came from and the date it was said. Nothing renders without both.
That shape is the whole product. This page uses it too — every number below carries the source it came from.
Veil API
One key for every major model, at each maker's own price
Call Claude, GPT, Gemini and more through one API, from Iran, with your existing code. Pay in Toman; see the exact cost of every request.
- Each maker's list price per token, with no markup. Veil earns a published 5% top-up fee.
- Works unchanged with the OpenAI, Anthropic and Google SDKs, Claude Code, Cursor and more.
- Veil never keeps your prompts (a batch you send is the one exception, encrypted until you collect its results). You choose which providers may see them.
- Pay in Toman at a published daily rate.
What is in the graph right now
- 0
- entities
- 0
- sourced facts
- 0
- independent sources
- 0
- typed relations
- 0
- have earned a page
Behind the veil
An entity gets a page only once it clears the publication floor. Most have not. A reference site that publishes a page per name collects thousands that each restate one fact from one source — veil would rather show you nothing.
Work nobody else has done
Persian tokenizer efficiency
A Persian speaker pays between 1.14× and 4.41× an English speaker's rate for the same meaning. Measured across 19 tokenizers, exact counts, reproducible from published files.
The FA/EN terminology lexicon
0 terms, each marked for whether the English word stays English. The synthesis pipeline obeys it.
The Iran AI access matrix
Which providers answer from Iran, by what route, reported by people who tried. An observation, never a guide.