TPSTokens Per Second

Nemotron 3 Content Safety 4B

NVIDIA🇺🇸 United States
active
Compare with other models →
Context window128K tokens

Version History

3.0major

Initial release of Nemotron 3 Content Safety 4B, featuring multimodal and multilingual capabilities trained on culturally-aligned safety data. Achieves 84% F1 accuracy on harmful-content benchmarks.

Coverage

model releaseNVIDIA

NVIDIA releases Nemotron 3 Content Safety 4B for multimodal, multilingual moderation

NVIDIA released Nemotron 3 Content Safety 4B, an open-source multimodal safety model designed to moderate content across text, images, and multiple languages. Built on Gemma-3 4B-IT with a 128K context window, the model achieved 84% average accuracy on multimodal safety benchmarks and supports over 140 languages through culturally-aware training data.

2 min read
content-moderationsafety-modelmultimodal