Unfragile

For text moderation, the source pages show few direct matches: Replicate hosts an open-source image content-moderation model 17, and ModerateContent.com offers a realtime image moderation API that blocks nudity 28. Neither is shown to moderate text, and the other products in the list do not show a text-moderation function on their pages.

A question back: Do you need to moderate text, images or both, and on which platform or stack will it run?

1
Replicatereplicate.comverified and signed
Replicate runs the kojott/content-moderation-vision model, which analyzes visual content and returns structured safety scores 17; it covers images rather than text.
Website Docs Pricing Get started
for agents: kojott/content-moderation-vision
2
ModerateContent.com Image Moderation APImoderatecontent.comverified and signed
A realtime API that moderates images for offensive content, with its GET / action blocking images with nudity 28; it does not show text moderation.
Website Docs Pricing
official:API · Image Moderation
for agents: GET /
3
Unprofaneunprofane.comwebsite read, tools not yet checked
A profanity API for checking text, with contexts, policies and evasion tags, and an MCP server for agents 5; its action record is not read.
Website Docs Pricing Get started
official:MCP server · Unprofane

Sources

1Replicate - Run AI with an APIreplicate.com · read 2026-10-29
2ModerateContent.com - Realtime Image Moderation APImoderatecontent.com · read 2026-10-29
5Documentation | Unprofane docsunprofane.com · read 2026-10-29
7kojott/content-moderation-vision: declares "kojott/content-moderation-vision" (the record we read and signed)unfragile.ai · read 2026-10-24
8Image Moderation: declares "GET /" (the record we read and signed)unfragile.ai · read 2026-10-24

For agents: the 10 actions behind this answer

kojott/content-moderation-vision · kojott/content-moderation-vision verified and signed
AI-powered content moderation for images using MiniCPM-V-2.6 - analyzes visual content and returns structured safety scores with detailed classifications
ditto--ai/qwen3guard-gen-4b · ditto--ai/qwen3guard-gen-4b verified and signed
A 4B-parameter safety and content moderation model that classifies user prompts and assistant responses as Safe, Unsafe, or Controversial with fine-grained category labels and refusal detection. Supports 119 languages.
fofr/prompt-classifier · fofr/prompt-classifier verified and signed
Determines the toxicity of text to image prompts, llama-13b fine-tune. [SAFETY_RANKING] between 0 (safe) and 10 (toxic)
meta/meta-llama-guard-2-8b · meta/meta-llama-guard-2-8b verified and signed
A llama-3 based moderation and safeguarding language model
adirik/grounding-dino · adirik/grounding-dino verified and signed
Detect everything with language!
GET / · Image Moderation verified and signed
Blocks images with nudity
openai/whisper · openai/whisper verified and signed
Convert speech in audio to text
· Lenz website read, tools not yet checked
Audit-grade fact-checking API for AI-drafted text: verdicts from a five-stage adversarial pipeline with a full citation trail.
zf-kbot/comfyui-text-unblur · zf-kbot/comfyui-text-unblur verified and signed
abiruyt/text-extract-ocr · abiruyt/text-extract-ocr verified and signed
A simple OCR Model that can easily extract text from an image.

Labels are the instruments' verdict and are never changed by ranking. Agents and robots get this same answer as JSON at /api/v1/find?q=…&answer=1 and through the MCP server.