Detoxify
Detoxify is a free software library that provides trained models and code to classify toxic comments.
What Detoxify is
Detoxify is a free software library that provides trained models and code to classify toxic comments. Its models were developed for three Jigsaw challenges using Wikipedia Comments and Civil Comments data. It offers original, unbiased, and multilingual models based on BERT, RoBERTa, and XLM-RoBERTa, with outputs that can include toxicity, threats, obscenity, insults, identity attacks, and explicit sexual content, depending on the model. The multilingual model is intended for English, French, Spanish, Italian, Portuguese, Turkish, and Russian. Developers can install the library with pip and use Python to predict labels for one string or a list; inference supports CPU by default and accepts a PyTorch device such as CUDA. The project identifies research, careful fine-tuning, and helping moderators flag harmful content faster as intended uses. It warns that profanity-associated words may be classified as toxic regardless of tone or intent, which can bias results against vulnerable minority groups. Detoxify is self-hosted, distributed under Apache-2.0, and its API requires an access request by email.
Good
- Free and self-hosted
- Three model variants available
- Multilingual model covers seven languages
- Python inference accepts one or multiple strings
- Apache-2.0 license
Watch out
- Profanity may be flagged regardless of intent
- Training requires Kaggle API access
- API access requires an email request
- No image, audio, or video moderation
Who it suits
It suits developers and researchers seeking self-hosted text classification for research, fine-tuning, or moderator triage. Its listed purpose and language support may fit projects that need comment classification rather than moderation of other media.
Verdict
Detoxify provides text-classification models and code that can be run on infrastructure you manage. Its documented bias warning matters when interpreting classifications, especially for profanity and tone.
Plans and prices
From Detoxify's own pricing page, checked 8 Oct 2026.
| Plan | Price | What it gives you |
|---|---|---|
| Open-source library | Free | Trained models and code for toxic-comment classification; API access by email |
The facts
Each one read from Detoxify's own pages, with the date it was checked.
What it does
- API access
- The repository says to email [email protected] to request access to its API.checked 8 Oct 2026
- Bias limitation
- The README warns that profanity-associated words may be classified as toxic regardless of tone or intent, potentially creating bias toward vulnerable minority groups.checked 8 Oct 2026
- Categories
- Model outputs include toxicity-related categories such as threats, obscenity, insults, identity attacks, and explicit sexual content, depending on the model.checked 8 Oct 2026
- Dependencies
- Inference uses Hugging Face Transformers and PyTorch Lightning, while training also requires the Kaggle API to download data.checked 8 Oct 2026
- Install
- The library can be installed with pip and used from Python to predict labels for one string or a list of strings.checked 8 Oct 2026
- Languages
- The multilingual model is intended for English, French, Spanish, Italian, Portuguese, Turkish, and Russian.checked 8 Oct 2026
- License
- The project is distributed under the Apache-2.0 license.checked 8 Oct 2026
- Models
- The library offers original, unbiased, and multilingual models based on BERT, RoBERTa, and XLM-RoBERTa, respectively.checked 8 Oct 2026
- Purpose
- Detoxify provides trained models and code to classify toxic comments from three Jigsaw challenges.checked 8 Oct 2026
- Runtime
- Inference supports CPU by default and accepts a PyTorch device such as CUDA.checked 8 Oct 2026
- Training data
- The models were trained for the 2018, 2019, and 2020 Jigsaw toxic-comment challenges using Wikipedia Comments and Civil Comments data.checked 8 Oct 2026
- Use
- The project describes its intended uses as research, fine-tuning on carefully constructed representative datasets, and helping moderators flag harmful content faster.checked 8 Oct 2026
Alternatives to Detoxify
- 1
Google Cloud Vision API
Free planWebAPILinuxmacOS7.9$0.60 usage-based - 2
Sightengine
Free planWebAPISelf-hosted7.5From $29 a month - 3
WebPurify
Free planAndroidiOSWebAPI7.4From $5 a month - 4
Amazon Rekognition Content Moderation
Free planWebAPI7.3From $0.10 a month - 5
Tisane API
Free planWebAPILinuxmacOS7.1From $99 a month - 6
Alibaba Cloud Content Moderation
iOSWebAPI7.0$0.09 usage-based