Linda-Stylo-Clean 1.0
Prefer a Windows app to a script? Linda Detector checks texts on your own PC and is free for personal, non-commercial use.
Windows application 2.0.6.2 β 17 October 2026
Download the release Β· Release notes Β· Try the online demo
Clear warnings before analysis, safer document handling and model-folder moves, responsive Loupe cancellation and redacted diagnostic metadata. Prefer at least 300 words of continuous prose; lists, forms and slide text may be unsuitable. Model package 1.3.4, tokenizers and thresholds are unchanged; short AI inserts can still be missed and results do not prove authorship.
The hosted demo sends text to Modal servers: 12 successful checks with a shared-network daily limit, up to 6,000 characters, TXT/MD/DOCX. Do not submit confidential text. The desktop app processes text locally and has no demo quota.
Stylometric AI-text detector for English (~250 style features + hashed n-grams -> LightGBM). CPU only, milliseconds per text, no GPU, no cloud. Trained only on data that pass a strict clean-provenance filter (open/permitted sources; AI texts generated by open-weight models with permissive terms). See DATA_BOM.md.
License: free for research / non-commercial use (CC BY-NC 4.0). Commercial use is paid β see COMMERCIAL_LICENSE.md, contact lindapro.support@proton.me.
pip install -r requirements.txt
python -m linda_stylo your_text.txt
from linda_stylo import LindaStylo
print(LindaStylo().detect(["text ..."])) # score + verdict (ai above the 1%-false-positive threshold, uncertain above 5%)
Measured (Chicago Booth benchmark, 1% threshold on the benchmark's humans): plain AI 97.7%, after StealthGPT 54.9%; false flags on school ELL essays 2.5%, TOEFL 2.2%. It is weaker than Linda Assay (formerly Linda-Pro), which adds two transformer voters, and it misses polished essays by modern models. Use it when data lineage, CPU-only operation or speed matter more than peak accuracy. English only. Not the sole basis for decisions about people. Contact: lindapro.support@proton.me
Documentation alignment β 17 October 2026
Current Windows app: 2.0.6.2; the combined update feed provides model package 1.3.4. This date describes the application/feed, not a new release of every individual model in this repository. See trust, benchmark brief and root LICENSING.md. Historical metrics retain their original versions and denominators.
Current deployment note: Windows Assay 2.0.6.2 uses FP32 stable DirectML (8 GB VRAM recommended; 4 GB unvalidated), Loupe uses CPU. Desktop history/update/licence behavior differs from the CLI and hosted demos. Historical speed and accuracy values must retain their original versions, modes and populations.
Free 14-day commercial evaluation
Free commercial evaluation lasts 14 calendar days from first launch, without a paid key, registration or automatic billing. This is a licence condition; the app does not enforce a timer. Continued commercial use then needs a paid licence. Personal non-commercial use stays free without a time limit. See COMMERCIAL_LICENSE.md for the full grant.