Persian LLM Reference
Local path: ~/Desktop/PERSIAN-LLM-REFERENCE/ — this folder is the git repo root for GitHub.
مرجع مدلهای زبانی فارسی
The open atlas for Persian (Farsi) language models — 72 sourced entries (v0.13.1) with verification gates, decision presets, gap mapping, normalized measured receipts, and bilingual JSON. Works alongside Awesome-Persian-LLM, MIZAN, PartAI, and the teams who build these models.
Community-maintained · sourced · machine-readable · online-first.
Related evaluation kit (same author): agent-security-bench — Agent Security Evaluation Kit. Walkthrough: docs/SCORE_AGENT_WITH_ASB.md.
Stats (v0.13.1)
| Count | |
|---|---|
| Total entries | 72 |
| Verified | 47 |
| Measured | 25 |
| Gap-tagged rows | 72 |
| Stable API | /api/v1/reference.json |
| PyPI | persian-llm-reference 0.13.1 |
| About | /about/ |
How PLR fits the ecosystem
| Resource | What it offers | What PLR adds alongside |
|---|---|---|
| Awesome Persian LLM | Curated discovery links | Structured records + verification gates |
| Leaderboards (MIZAN, PartAI) | Benchmark scores | Links to leaderboards inside one manifest |
| Hugging Face search | Model hosting + cards | Persian taxonomy · corpus · script fields |
| Papers & repos | Primary artifacts | Indexed rows with asOf dates and cite URLs |
Upstream discovery: Awesome-Persian-LLM — verified community index awesome-persian-llm · ECOSYSTEM.md. Bibliographic link only — not an endorsement or partnership.
Live surfaces (after deploy)
| Surface | URL |
|---|---|
| Atlas UI | https://sinakazemnezhad.github.io/persian-llm-reference |
| PyPI (Python client) | https://pypi.org/project/persian-llm-reference/ |
| Manifest JSON | /data/reference-manifest.json |
| Stable API v1 | /api/v1/reference.json (alias of canonical manifest) |
| Raw manifest (cite in papers) | https://raw.githubusercontent.com/sinakazemnezhad/persian-llm-reference/main/data/reference-manifest.json |
| Source radar (field gaps + forecast) | /data/source-radar.json |
| Agent discovery | /.well-known/persian-llm-reference.json |
Repository layout
| Path | Role |
|---|---|
data/reference-manifest.json |
SSOT — all models, datasets, benchmarks |
workspace/public/ |
Static site (GitHub Pages root after build) |
governance/MANIFEST_LAW.md |
Taxonomy + verification gates |
governance/METHODOLOGY.md |
Trust, axes, cite policy |
governance/POSITIONING.md |
What PLR is — alongside other references |
governance/ROADMAP.md |
Full phased plan — meaningful · usable · productive · smart |
docs/PYPI_TRUSTED_PUBLISHING_SETUP.md |
PyPI OIDC publish (sourcea-boot pattern) |
pyproject.toml |
Python package · pip install persian-llm-reference |
GitHub setup (required)
- Repo: github.com/sinakazemnezhad/persian-llm-reference
- Enable GitHub Pages → source: GitHub Actions
- Push
mainbranch - Site:
https://sinakazemnezhad.github.io/persian-llm-reference
Python client (PyPI)
Same pattern as sourcea-boot · pypi.org/project/sourcea-boot.
pip install persian-llm-reference
plr stats
plr get persianmind-v1
plr cite
plr validate
Editable install from repo root:
cd workspace && node scripts/sync-python-manifest.mjs
pip install -e .
- Python client: MIT (
LICENSE-MIT) - Registry metadata: CC-BY-4.0 (
LICENSE) - Publish: GitHub Release →
publish-pypiworkflow (trusted publishing). Seedocs/PYPI_TRUSTED_PUBLISHING_SETUP.md.
Build & verify
cd workspace
npm run build # validate + copy manifest to public/
npm run e2e # local server smoke
Verification gates
indexed → verified → measured → corpus-documented → literary
No invented scores. null until a primary source exists.
How to cite
PLR is a neutral evidence registry — cite the snapshot you used, not a ranking.
- Machine-readable: CITATION.cff (version, date-released, abstract, repository URL).
- BibTeX (copy for papers):
@misc{persian_llm_reference,
title={Persian LLM Reference — Open Atlas},
author={{Persian LLM Reference maintainers}},
year={2026},
version={0.13.0},
url={https://sinakazemnezhad.github.io/persian-llm-reference/},
howpublished={\\url{https://raw.githubusercontent.com/sinakazemnezhad/persian-llm-reference/main/data/reference-manifest.json}},
note={Check manifest generatedAt for snapshot date; measured rows include benchmark, metric, source, and conditions}
}
- API consumers: pin
GET /api/v1/reference.jsonor the raw manifest URL — same schema, same evidence semantics.
Also see CITATION.cff for Citation File Format metadata.
Contributing
One PR = one entry + primary source URL. See governance/CONTRIBUTING.md.
License
Manifest metadata and docs: CC-BY-4.0. Each model entry retains its own license.
Metadata
Release files for persian-llm-reference 0.13.1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| persian_llm_reference-0.13.1.tar.gz | 25.7 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| persian_llm_reference-0.13.1-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 50.1 kB
Release files / persian_llm_reference-0.13.1.tar.gz
| Download URL | persian_llm_reference-0.13.1.tar.gz |
|---|---|
| Size | 25.7 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
d4046862b73edba27ed3e02296d8ed9899817243be9125abbd2d1a457fe8215d
|
|
BLAKE2b-256 checksum How to use checksums |
0bef8b47bb3f98c676ea8d5f605747bab6b85397a9013f86fd153aece4d43064
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Aug 16, 2026.
Transparency logRelease files / persian_llm_reference-0.13.1-py3-none-any.whl
| Download URL | persian_llm_reference-0.13.1-py3-none-any.whl |
|---|---|
| Size | 24.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
28ea8906f483f5d702f1cc564fd886c9fac57ba05a505097b5da6c712d2e3a30
|
|
BLAKE2b-256 checksum How to use checksums |
f9c7222eb9041c82c6615ba374462221e49535b7b5934751f2e6f7340198f9f3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Aug 16, 2026.
Transparency log