Stop harmful content before your users see it
A single inappropriate photo can destroy trust built over years. modwall automatically checks every upload — images and text — protecting your platform's reputation and helping you meet DSA requirements. Images are processed in memory and never stored.
No card required · 5-minute setup · Hosted in the EU · Przelewy24 payments, VAT invoice
Under the DSA, moderation is no longer optional
Every platform with user-generated content — a marketplace, forum, classifieds site or social app — is responsible for what its users publish. Building and running your own 24/7 moderation is a full-time job a small team doesn't want. modwall takes it off your plate: one integration, zero maintenance on your side.
Try it on your own photo — no sign-up
Drag in a photo and see the decision exactly as your backend will see it. The image goes to memory only for the duration of scoring — we do not store it.
The demo scores three categories: NSFW (nudity / adult content), weapons and violence/gore (beta). Offensive gestures are not detected; text toxicity is handled by a separate model (/v1/moderate-text). In the service you enable and tune each category separately, per site.
All categories included in your plan
With many providers, each detection category is a separate model eating operations from your paid pool. With us, a single request scores every category you enable — and you enable them per site (moderation profiles): an adult site without the nudity filter, a gaming site without the weapons filter. You know your context best.
🔞 NSFW — nudity and adult content
The backbone of the service since day one. A specialized classifier, a raw score and your own threshold — from a lenient forum to an app for schools.
🔫 Weapons (new)
Handguns, rifles, knives in hand. On our validation set: 99% precision at the block threshold (about 7 false positives per 1000 ordinary photos), and the review band raises recall to ~98% — borderline cases go to your queue, not to your platform.
🩸 Violence / gore (beta)
Fights, injuries, graphic scenes. This category is in beta — start in monitor mode and calibrate the threshold on your own traffic before you enable blocking.
💬 Text toxicity
Insults, threats, hate speech — a multilingual model that understands Polish. The same API pattern (/v1/moderate-text) and a shared request quota.
Enable categories in your dashboard Compare with other providers
5 minutes to your first request
You get a key, copy three lines and see a working response.
curl -X POST https://modwall.dev/v1/moderate \
-H "Authorization: Bearer sk_live_..." \
-F "image=@zdjecie.jpg"
# => {"scores":{"nsfw":0.97,"safe":0.03},"decision":"block",
# "recommended_action":"block","mode":"enforce","threshold":0.8}
You always get the raw score — you set the threshold and mode for your own context. Text is checked with the same pattern via /v1/moderate-text. A ready-made website widget and examples in Python, Node.js and PHP are in the documentation.
Not writing code? Ready-made plugins
Running a forum, a gallery or a WordPress site — without a development team? A plugin wires in moderation in minutes: new images are scanned, suspicious ones go to a queue you already know, and the decision always stays with a human.
Chevereto
Auto-approval of safe uploads in your gallery. Replaces the disappearing ModerateContent integration.
WordPress
Media library scanning plus quarantine with a decision panel in the admin dashboard.
phpBB
Posts with flagged attachments land in the native approval queue (MCP).
Zero risk to start
Plug it in and just watch — monitor mode blocks nothing
modwall scores your real traffic and shows in the dashboard what it would flag — with no change to your users' experience. You watch the results on your own data and decide when to turn on enforcement. It's the safest way to try moderation in practice.
Why modwall
Compliance-grade
Designed for the DSA and GDPR. An immutable, hash-chained audit log of every decision — a trail for the regulator that a wrapper doesn't have.
Privacy / EU
The image never leaves the server. In-memory processing, zero storage, EU hosting. A DPA ready to sign as standard.
Self-host
Our container on your infrastructure — the image never leaves it. For sensitive sectors and legal requirements.
Predictable cost
Simple plans with no surprises — you pay for a problem taken off your hands, not for tokens. Start at 0 zł, VAT invoice.
5-minute setup
A server-side API and a drop-in widget. Zero maintenance on your side — uptime, scaling and the model are our job.
You stay in control
Monitor / review / enforce modes, per-account thresholds, a human-in-the-loop queue and webhooks. modwall signals — you decide.
How it works
1 · Upload
Your application sends the image to /v1/moderate at upload time — from your backend or via the widget.
2 · Scoring
The model returns a raw probability score for NSFW content (nudity / adult content). The image is gone from memory right after inference.
3 · Decision
According to your policy: allow, hold for review (HITL) or block. Everything goes into the audit log.
Who it's for
For every platform where users upload photos — and you are responsible for what appears on it.
Marketplace
Przefiltruj zdjęcia ofert i awatary sprzedających, zanim trafią do katalogu — bez budowania własnego pipeline'u moderacji.
Classifieds sites
Tysiące zdjęć dziennie ocenianych automatycznie, z kolejką do ręcznego przeglądu tylko tam, gdzie naprawdę trzeba.
Forums and communities
Skanuj załączniki i awatary przy wgrywaniu — moderatorzy dostają tylko przypadki graniczne, nie całą skrzynkę.
Dating
Zdjęcia profilowe sprawdzone przed publikacją — chroni użytkowników i reputację aplikacji od pierwszego uploadu.
Comments and UGC
Obrazki i tekst w komentarzach oraz recenzjach pod kontrolą — bez budowania własnego pipeline'u moderacji.
Apps with avatars
Wszędzie, gdzie użytkownik wgrywa zdjęcie profilowe — wpięcie widgetem w kilka minut, bez zmian w backendzie.
“But the model is open-source — why pay?”
Because “I could build it” is not the same as “I have time for it”. Wrapping a model is a weekend — running it in production is for life: uptime, scaling, monitoring, the 3 a.m. on-call, model updates. You are not paying for the model, but for what you don't want to do yourself:
Reliability
It just works. You maintain nothing.
Compliance
DPA, processing in the EU, an audit log — the paperwork a DSA regulator expects.
Convenience
Onboarding in 5 minutes, then zero maintenance on your side.
We sell the last 90% — reliability, compliance, and peace of mind — not the first 10%, the model.
Built by someone who has done serious moderation
modwall is built by the operator of a community of ~300,000 users. We know the pain of moderation from the side that hurts the most — yours. That's why we don't sell yet another model, but reliability, compliance and peace of mind.
Plans
You pay for the result, not the model. You pick a scope — images only, text only, or both bundled — and pay only for that. A limit per request, reset every month. Net prices (+23% VAT). No card to start.
Self-host / on-premise
Our container on your infrastructure — the image never leaves it. The highest level of privacy, annual contract.
Get your key in 2 minutes
Create an account, get 1,000 free requests a month and wire in moderation today.
Create a free accountmodwall is a tool that assists moderation — the final decision and responsibility remain with the customer. CSAM material is outside the scope of the service.