A research team from the U.S.-based AI startup Perplexity has developed a modified version of the Chinese AI language model DeepSeek-R1 and released it as open source.
This new model, called “R1 1776”, has been adjusted to operate without the original censorship restrictions. The original DeepSeek-R1 adhered to Beijing’s censorship policies and refused to answer certain sensitive topics, such as Taiwan, the Uyghur minority in China, and the events of Tiananmen Square in 1989.
By applying special retraining techniques, the researchers successfully removed these limitations, allowing the model to respond freely to such topics while maintaining its original strengths in mathematics and logical reasoning.
The model is now available via the Hugging Face repository and can also be accessed through the Sonar API. The source code has been released under an open-source license, making it accessible for researchers and developers worldwide.
