cff-version: 1.2.0
message: "If you cite DotCheck engines or measured holdout gates, use the citation below."
title: "DotCheck AI-likeness engines (inhouse@14, inhouse-text@12, inhouse-video@4)"
authors:
  - name: "DotCheck"
url: "https://dotcheck.ai"
repository-code: "https://huggingface.co/DotCheck"
license: "Apache-2.0"
date-released: "2026-08-25"
abstract: >-
  DotCheck serves binary AI-likeness scores for images (wire inhouse@14 /
  Vermeer@14.2, pair stack: max-side 256 transport + native 224 center),
  text (inhouse-text@12 / Valla@12.2, eight language heads), and video stamps
  (inhouse-video@4 / Muybridge@4.4, same pair views then stamp-cap max)
  using commercial-clean SigLIP2 spines plus Apache-2.0 heads. Live heads
  are published on Hugging Face; everyday product scoring runs through
  DotCheck.
keywords:
  - ai-detection
  - image-classification
  - text-classification
  - video
  - siglip2
preferred-citation:
  type: soft
  authors:
    - name: "DotCheck"
  title: "DotCheck engines (inhouse@14, inhouse-text@12, inhouse-video@4)"
  url: "https://dotcheck.ai/docs"
  year: 2026