NSFW JS
Verified Oct 15, 2026 ActiveMIT-licensed JavaScript model that sorts images into porn, hentai, sexy, drawing or neutral in-browser
Listed on EffectHub ·
Pricing
- Pricing model
- 100% Free
MIT-licensed open source with no paid tier and no hosted API. The only cost is the compute of running the model yourself.
Use tool ↗About NSFW JS
NSFWJS is an open-source classifier rather than a service: you install the library, it loads a TensorFlow.js model, and images are scored locally into five buckets (drawing, neutral, sexy, hentai, porn) so nothing is uploaded to a moderation vendor. It runs in the browser, under Node with the tfjs-node backend, and in React Native. The maintainers report roughly 90 to 93 percent accuracy, and you can self-host the weights to cut bundle size or swap in your own model. It is maintained by Infinite Red, a US development consultancy, under the MIT licence; nsfwjs.com is the live demo. There is no hosted API and nothing to pay.
- Platforms
- Web
- API
- Not confirmed
- Last verified
- Oct 15, 2026
Verified sources
- GitHub repository infinitered/nsfwjs: MIT licence, not archived, latest release v4.4.0 published 4 August 2026, repository created February 2019, homepage nsfwjs.com (15 October 2026)github.com ↗
- nsfwjs.com serves the client-side demo, described as a JavaScript library that identifies unseemly images in the browser (15 October 2026)nsfwjs.com ↗
Is NSFW JS legit?
We don’t label tools safe or unsafe. Below are the records we checked ourselves, each with the date we read it.
- Domain registered
- Feb 14, 2019 (7 years ago)Registry record (RDAP), registrar Gandi SAS, looked up Sep 6, 2026
- TLS certificate
- Issued by Let's Encrypt, valid until Nov 29, 2026Seen in our own connection to the site on Oct 15, 2026
- Last checked
- Oct 15, 2026The site answered
- On the market since
- 2019The vendor's own material, read by us
- Operated by
- Infinite Red, Inc., United StatesOur research, card verified Oct 15, 2026
Pros & cons of NSFW JS
The right choice when you need a first-pass nudity filter and cannot or will not send user images to a third party: it runs client-side, costs nothing, and the licence permits commercial use. Accuracy around 90 percent makes it a triage layer, not a compliance control, so expect false flags on swimwear and misses on borderline material. There is no dashboard, no policy tuning and no support contract.
Pros
- Runs entirely client-side, so images never leave the device
- MIT licence, free for commercial use
- Works in the browser, in Node via tfjs-node, and in React Native
- Model weights can be self-hosted or replaced with your own
Cons
- Roughly 90 percent accuracy, which is triage-grade rather than compliance-grade
- Library only: no hosted API, dashboard or support contract
- Five fixed categories with no policy configuration
- Loading the model adds meaningful weight to a web bundle
Best for
- client-side pre-filtering before an upload leaves the browser
- privacy-sensitive apps that cannot ship user images to a vendor
- teams wanting a free first pass ahead of paid moderation
Not for
- legal or regulatory content-safety obligations
- anyone who needs a hosted API with an SLA
Watch out: It only detects nudity-adjacent categories. Violence, weapons, self-harm and CSAM detection are outside what this model does.