---
language:
- fa
- en
license: cc-by-4.0
task_categories:
- text-generation
- question-answering
tags:
- persian
- farsi
- llm
- benchmark
- registry
pretty_name: Persian LLM Reference Manifest
size_categories:
- n<1K
---

# Persian LLM Reference — manifest snapshot

Bilingual, receipt-gated registry of Persian (Farsi) language models, datasets, benchmarks, and leaderboards.

**This Hub dataset mirrors a release snapshot. It is not a competing source of truth.**

## Canonical surfaces (use these)

| Role | Surface |
|------|---------|
| **Canonical reference (machine)** | [API v1 `/api/v1/reference.json`](https://sinakazemnezhad.github.io/persian-llm-reference/api/v1/reference.json) |
| **Human interface** | [PLR Atlas](https://sinakazemnezhad.github.io/persian-llm-reference/) |
| **GitHub SSOT** | [manifest on `main`](https://github.com/sinakazemnezhad/persian-llm-reference/blob/main/data/reference-manifest.json) |
| **Versioned release** | [GitHub Releases](https://github.com/sinakazemnezhad/persian-llm-reference/releases) |
| **HF Dataset (mirror)** | [Noetfield/persian-llm-reference](https://huggingface.co/datasets/Noetfield/persian-llm-reference) |

## Files in this dataset

- `reference-manifest.json` — manifest at upload time. Check `version` and `generatedAt` inside.

## Cite

[CITATION.cff](https://github.com/sinakazemnezhad/persian-llm-reference/blob/main/CITATION.cff) · pin `version` + `generatedAt`.

```text
Claims map to public evidence. Unsupported claims remain explicitly unverified.
```